Staten Island University Hospital (SIUH) petitions for review of an April 29,1993 order of the National Labor Relations Board (NLRB) holding SIUH to have committed unfair labor practices. The NLRB cross-petitions for enforcement of that order. The NLRB found that SIUH wrongfully refused to bargain with a duly elected union representing registered nurses (RNs) at one of SIUH’s two sites. SIUH asserted then, as it does now, that its refusal was justified. According to SIUH, the site’s RNs should not have been recognized as a separate bargaining unit, for two reasons: first, that the RNs were accreted in a merger of the two sites, and second, that the RNs lack a sufficient community of interests to constitute an appropriate bargaining unit. The NLRB rejected those assertions. The Federation of Nurses, United Federation of Teachers, AFL-CIO (UFT) intervenes in support of the NLRB. The New York State Nurses Association (NYSNA) intervenes in support of SIUH. We deny SIUH’s petition for review, and grant the NLRB’s cross-petition for enforcement. The record supports the NLRB’s conclusions that no accretion took place, and that the RNs at each site constitute an appropriate separate bargaining unit.
*452I.
The following undisputed facts were established before the NLRB. Staten Island Hospital and Richmond Memorial Hospital were located about eight miles apart on Staten Island, New York. In 1987, they merged their boards of directors and chief executive offices. The resulting institution was named Community Health Systems of Staten Island, Inc. The institution did not achieve the economies of scale that motivated the upper-level merger, and in 1989, the board of directors ordered that all other operations be merged as fully as possible. In 1990, the institution was renamed Staten Island University Hospital. The former Staten Island Hospital is now the “north site” of SIUH, and the former Richmond Memorial Hospital is the “south site” of SIUH.
SIUH employs about 3000 people: over 2300 at the north site and about 650 at the south site. According to October 1991 figures, RNs number 909. They are categorized as full-time (37.5 hours per week), half-staff (18.75 to 37.5 hours per week), per diem (7.5 to 18.75 hours per week), and casual (0-7.5 hours per week). North site RNs are 44% full-time, 26% half-staff, 19% per diem and 11% casual. NYSNA represents 679 RNs, most of whom work at the north site, but 80 of whom work at other locations, including the south site’s Alcohol Detoxification Unit.1 The remaining 230 RNs work at the south site. South site RNs are 31% full-time, 29% half-staff, 34% per diem, none casual, and 6% “part-time” (7.5 to 37.5 hours per week, a category unique to the south site).
Organization of the institution is more integrated as a result of the second merger. A single medical director is responsible for both sites. Single department heads administer all clinical and operating departments, nursing, finance, human resources, purchasing, planning, risk management, materials management, management information, dietary, housekeeping and environmental services. A single institutional budget has been adopted. Uniform employee policies have been implemented. A common internal telephone system was installed. A uniform correspondence policy was adopted: the same address is used for both sites; forms were consolidated; references to Staten Island Hospital and Richmond Memorial Hospital were eliminated; and references to the north and south sites are minimized. The same ambulance and paramedic teams serve both sites. A shuttle bus transports supplies and mail between sites five times daily. A single public relations office handles advertising, and does not differentiate the two sites in its work. And the Joint Commission on Accreditation of Healthcare Organizations (“JCA-HO”), which qualifies hospitals for Medicare participation, issued a single accreditation for the two facilities.
Medical functions are more integrated as a result of the second merger, although in many respects the sites still provide services as independent hospitals. Basic medical functions are performed at both sites: e.g., medical/surgical, emergency, ICU, CCU, he-modialysis, operating and recovery, ambulatory, laboratory, pharmacy, nuclear medicine and physiological. Certain specialized functions are performed at only one site: e.g., MRI, EEG, obstetrics, pediatrics, cardiac catheterization, in-patient rehabilitation, substance abuse, and certain laboratory services.
The human resources department is headed by a single senior vice president, and has ultimate authority over employment matters for all employees, including RNs. These matters include interviewing, hiring and discipline, orientation and continuing education, dress codes and working hours, salaries and benefits, employment forms, and records storage. Yet the central office’s involvement in personnel decisions is limited. In hiring, for example, the central office only initially screens applicants and performs physical examinations. Site-specific unit managers and supervisors select the persons to hire, and the central office accepts those selections.
The SIUH nursing plan describes its goal to be a “decentralized management structure.” General matters (e.g., job duties, *453medical record forms, quality assurance, infection control, medications, and dietary, therapeutic and social services) are uniform throughout both sites. The nursing department is headed by a senior vice president and a vice president for nursing operations, each of whom has an office at both sites. Next in the chain of command are assistant directors of nursing (“ADNs”), who supervise multiple hospital units. ADNs ordinarily remain at a single site, but four of them travel between sites to cover for absent ADNs. ADNs hold joint meetings. ADNs are supplemented by administrative assistant directors of nursing (“administrative ADNs”) who remain at a single site. Finally, nurse managers (“NMs”) and nursing care coordinators (“NCCs”) supervise at the unit level. NMs and NCCs administer day-to-day operations. NMs supervise single hospital units — e.g., intensive care or emergency— which comprise about 25-40 nursing and non-nursing staff. NMs handle scheduling (including vacation and overtime), grievances not resolved by NCCs, unit budgets, and employment decisions. NMs hold joint and single-site meetings. Each NCC assigns and supervises about 9 RNs’ duties and schedules, prepares evaluations, reprimands RNs and initially handles grievances, and participates in employment decisions.
Hours and wages are substantially identical for RNs at both sites. Other aspects of compensation, however, differ significantly. The north and south sites differ in distribution of benefits between full-time and other RNs. Seniority governs numerous benefits, and north site half-staff RNs accrue seniority on a pro-rated basis relative to full-timers. South site half-staff and part-time RNs, by contrast, accrue seniority at a rate equal to south site full-timers’ rate. Health coverage is similarly provided on a pro-rated basis at the north site, but not at the south site.
Each site has its own labor relations office and employee relations specialist to handle grievances, answer employee inquiries and report to the central human resources department. The employee relations specialists handle third-step labor grievances, which go to arbitration if unresolved. The specialist at the south site answers to a central employee relations manager, who doubles as the employee relations specialist for the north site. RN grievances that do not implicate labor relations matters are handled by NCCs, with review authority in NMs, ADNs, and eventually vice-presidents.
RNs may voluntarily transfer from site to site, but in such cases they are treated as new hires. Transfers are in any event discouraged by SIUH and JCAHO policies. Involuntary transfers are opposed by RNs at both sites; they have never occurred. Job openings for each site are posted at the other site. From January 1990 through September 1991, 175 RNs were hired for the north site and 83 RNs were hired for the south site. Fifteen of the north site hires were transfers from the south site. Seven of the south site hires were transfers from the north site.
RNs fill temporary needs in other units at the same site, never at the other site. Inter-site transfer to fill a temporary need is against SIUH and JCAHO policies, as well as RNs’ wishes. But RNs from the north site teach JCAHO-required orientation and continuing education sessions at both sites, which sessions are attended by RNs from both sites.
Patients may voluntarily transfer between sites. In the first eight months of 1991, 230 of 18,500 patients (1.2%) transferred between sites: 170 from south to north and 60 from north to south. Patient transfers require intersite communication among RNs. In the first eight months of 1991, about 337 instances of communication occurred. These communications, however, do not differ from what occurs when patients transfer between SIUH and unrelated hospitals. Patient records are kept at the site from which a patient is discharged.
NYSNA has represented north site RNs since 1972; its most recent agreement ran for two years until the end of March 1993. The Richmond County Nurses Association (RCNA) began representing south site RNs in 1974; its most recent agreement was set to expire at the end of August 1990, but in January 1989, that agreement was extended to the end of October 1991.
*454In July 1991, UFT filed a petition before the NLRB to represent south site RNs. SIUH and NYSNA opposed the petition on the ground that the mergers had resulted in an accretion of south site RNs to the north site RN bargaining unit. Hearings were held before an NLRB hearing officer from August to November 1991. In March 1992, the NLRB Regional Director issued a decision that the south site RNs: (1) were not accreted to those at the north site by the merger; and (2) constituted an appropriate separate bargaining unit for election purposes. He directed that an election be held. SIUH and NYSNA requested NLRB review of the Regional Director’s decision, and a stay of the election. UFT opposed the request for review and stay. The NLRB affirmed the Regional Director’s decision by order dated July 29, 1992, and denied the request for a stay. Staten Island Univ. Hosp., 308 N.L.R.B. 58, 1992 WL 187613 (1992).
On September 18, 1992, an election was held among south site RNs. UFT won with 150 votes. NYSNA received 5 votes. RCNA received no votes. Three votes were cast for no representation. On October 5, 1992, the NLRB certified UFT as the south site RNs’ representative.
Later that month, UFT twice wrote SIUH and requested to bargain. SIUH responded both times by refusing to bargain. In November 1992, UFT filed an unfair labor practice charge against SIUH.
In December 1992, the NLRB Regional Director issued an unfair labor practice complaint. In March 1993, the NLRB General Counsel moved for summary judgment. SIUH requested a hearing to introduce two new items of evidence. First, SIUH sought to introduce a February 1993 survey and affidavit indicating that SIUH was the only metropolitan area acute care facility that was required to negotiate with two RN units in a single nursing department. Second, SIUH sought to introduce an affidavit of SIUH’s president describing how separate bargaining units impeded establishment of a traumatic brain injury unit. The brain injury unit was to be established at the south site, but like the alcohol detoxification unit, was to be staffed with experienced north site operating room nurses. SIUH’s president expressed fear that UFT, as the new representative of south site RNs, would block establishment of the brain injury unit. UFT had previously expressed displeasure with RCNA’s concession enabling NYSNA RNs to work in the south site alcohol detoxification unit.
On April 29, 1993, the NLRB denied SIUH’s hearing request and granted the General Counsel’s motion for summary judgment. Staten Island Univ. Hosp., 310 NLRB No. 207, 1993 WL 138838 (1993). The NLRB held that SIUH committed an unfair labor practice in refusing to bargain with UFT, and ordered SIUH to cease and desist from those practices, to bargain with UFT, and to post notices about the order.
II.
SIUH, joined by NYSNA, asserts that the north and south sites are too well-integrated to justify separate bargaining units. The NLRB and UFT maintain that the NLRB’s decisions were within its discretion. We agree with the NLRB and UFT that the NLRB permissibly exercised its discretion.
The NLRB made two findings supporting separation of the north and south site bargaining units. First, the NLRB found that the mergers did not result in an accretion of south site RNs to the north site bargaining unit. Second, the NLRB found that the south site RNs constitute an appropriate separate bargaining unit, and ordered an election.
The accretion and unit determination findings depend on the same basic concern: the degree to which employees in separate locations share a community of interests distinct from their interests as employees of the whole institution. The degree of shared interests is measured by eight factors: geographic proximity, similarity of skills and functions, similarity of employment conditions, centralization of administration, managerial and supervisory control, employee interchange, functional integration of the employer, and bargaining history. NLRB v. *455Stevens Ford, Inc., 773 F.2d 468, 473 (2d Cir.1985).
The primary difference between accretion and unit determination analyses is one of degree rather than kind. Accretion requires an overwhelming community of interests between a smaller group of employees and a larger unit, because accretion casts the smaller group, against its will, into the larger unit. Accretion therefore applies only if one group of employees has no identity distinct from the other. Local 144, Hotel, Hosp., Nursing Home & Allied Servs. Union v. NLRB, 9 F.3d 218, 223 (2d Cir.1993). Unit determination, by contrast, requires only a substantial community of interests among a group of employees to support casting them as a unit. Substantial communities of interest may be found for units of varying scope, and the NLRB enjoys discretion to select from those possible arrangements in reaching its unit determination. American Hosp. Ass’n v. NLRB, 499 U.S. 606, 610, 111 S.Ct. 1539, 1542, 113 L.Ed.2d 675 (1991). The determination thus requires selection of an appropriate unit, not the most appropriate unit, id., and the NLRB’s decision will stand unless arbitrary and unreasonable, Continental Ins. Co. v. NLRB, 409 F.2d 727, 728 (2d Cir.), cert. denied, 396 U.S. 902, 90 S.Ct. 215, 24 L.Ed.2d 178 (1969).
SIUH and NYSNA assert that the NLRB exceeded its discretion both in finding no accretion and in finding that the south site RNs constitute an appropriate bargaining unit for unit determination purposes. We disagree. The NLRB acted within its discretion. Our conclusion follows from our review of the factors for evaluating the degree of employees’ shared interests.
The geographic proximity factor neither compels nor precludes the NLRB’s decision. The sites are separated, but only by eight miles.
The skills and functions factor similarly offers little guidance. SIUH asserts that RNs use the same skills and perform the same basic functions at the two sites, regardless of the different functions performed in particular care units. The NLRB contends that each site has specialized units and functions, and therefore requires specific skills, that the other site does not. Although this factor might weigh slightly in favor of accretion and a single employer-wide unit, the NLRB’s contrary position has support.
Differences in employment conditions are a more significant factor, and lend firm support to the NLRB’s decision. North site and south site RNs have different seniority and benefits packages, not only in the options and levels of benefits, but in the treatment of full-time and other RNs. NYSNA treats north site half-staff RNs less favorably than RCNA and UFT treat south site half-staff and part-time RNs, which may help explain why south site RNs — a greater proportion of whom are half-staff or part-time rather than full-time— overwhelmingly rejected NYSNA (150 to 5) in the September 1992 election. SIUH asserts that wages are identical, and that all differences in other benefits and compensation are due to historical differences in representation. But the differences cannot be ignored. South site RNs’ interests remain divergent from north site RNs’ interests. Separate representation protects the south site RNs.
The next two factors — administrative centralization and common managerial and supervisory control — again present conflicting considerations. Nursing lends itself to central institutional administration, but also requires close site-specific supervision. Thus in this case, nursing operations are administratively centralized, but day-to-day supervision is performed by NCCs, NMs and assistant ADNs, all of whom remain in one unit or site. Higher-level administrators rarely come in contact with RNs or patients. Perhaps these operations are centralized as much as is safely possible in a multiple-site hospital. Surely an employer-wide nursing unit is permissible at a multiple-site hospital, despite the inevitability of some degree of site-specific supervision. On balance, however, these factors do not weigh heavily in either direction. Centralized administration favors an employer-wide unit, but site-specific supervision favors separate units.
The employee interchange factor offers clearer guidance. Little RN interchange occurs. Of 909 RNs, only 4 to 6 regularly work *456at both sites. Voluntary transfers account for a small percentage of hires, and there have been no involuntary transfers. SIUH and JCAHO policies discourage transfers, and RNs themselves oppose involuntary transfers. These considerations favor separate units.
We also find little functional integration that would necessitate a single bargaining unit.' Numerous functions traditionally performed at just one site have neither moved nor expanded, with the exception of the alcohol detoxification unit. The two sites depend on one another for certain services, but that is true of completely unrelated hospitals. And although the sites share service and technical employees (1200, all in a single bargaining unit), and integrated communications and transportation, we find those considerations collateral to the core functions of the hospital. The small amount of patient transfer (230 of -18,500, or 1.2%, during a typical period) bolsters the conclusion that the hospitals essentially function independently. The functional integration factor favors separate units.
Bargaining history lends additional support to separate units. SIUH bargained separately with NYSNA and RCNA after the initial corporate merger, and extended RCNA’s representation not long before formalizing the second and more comprehensive merger." The sites historically"have enjoyed separate representation. We understand SIUH’s contention that it accepted separate units only as long as necessary — that the last extension of RCNA’s representation was contractually required before the second merger effort, and that SIUH then attempted to consolidate the units as soon as it could. However, we believe that the balance of these considerations weighs in favor of separate units.
The independence and shared identity of south site RNs was dramatically confirmed by the September 1992 election in which UFT received 150 votes to NYSNA’s 5 votes. The south site RNs’ recent and overwhelming rejection of NYSNA shows that accretion would be particularly harsh, and lends support to the determination that they constitute an appropriate separate bargaining unit.
The NLRB’s decisions on accretion and unit determination are supported by substantial evidence, particularly regarding seniority and benefits, employee interchange, functional integration and bargaining history. The south site RNs retain a distinct identity. They consequently cannot be accreted, and may properly be designated as a separate unit.
SIUH asserts that the NLRB wrongly relied on the “single facility presumption,” which holds that employees in a separate facility (like the- south site) should be presumptively treated as a separate bargaining unit. SIUH cites our prior position that “the single-facility presumption is inapplicable in the health care context.” Long Island Jewish-Hillside Medical Ctr. v. NLRB, 685 F.2d 29, 34 (2d Cir.1982). The NLRB counters that our recent decision in Local 144, 9 F.3d 218, abrogates our prior position. We recognized in Local HI that the Supreme Court’s decision in American Hospital, "499 U.S. 606, 111 S.Ct. 1539, undercut our prior proscription of the single-facility presumption. We concluded that “we cannot say that the Board was bound to follow this Circuit’s preference for consolidating bargaining units in multi-site hospital situations.” Local 144, 9 F.3d at 225.
In fact, American Hospital gave the NLRB’s use of presumptions a more ringing endorsement than our Local 144 decision might suggest. The Court upheld an NLRB rule providing that, with only three exceptions, eight and only eight bargaining units shall be recognized at acute care hospitals. The petitioner • in American Hospital challenged the rule as “qualitatively different from ... prior rules[ ] which at most established rebuttable presumptions that certain units would be considered appropriate in certain circumstances.” 499 U.S. at 612, 111 S.Ct. at 1543. The Court upheld the even stricter rule — and thus left the mere rebuttable presumptions beyond dispute. We regard the single-facility presumption as the kind of rebuttable presumption that was beyond dispute in American Hospital. The NLRB is not required to exercise “standard-less discretion in each case.” Id. It has authority to “resolve certain issues of general *457applicability unless Congress clearly expresses an intent to withhold that authority.” Id. The NLRB has good reasons to use the single-facility presumption.
SIUH asserts that even if the single-facility presumption applies, it is overcome in this case for two reasons: first, that the community of interests (as measured by the factors discussed supra in connection with accretion) favors an institution-wide unit; and second, that separate units threaten SIUH with labor disputes, wage and benefit whipsawing, and inability efficiently to transfer RNs from site to site in reorganizations. We disagree. South site RNs share a substantial community of interests distinct from the interests they share with north site RNs.
The threat of work stoppage does not make separate units unreasonable in this ease. The NLRB is aware that recognition of multiple bargaining units threatens quantitatively greater labor disruptions. West Jersey Health Sys., 293 N.L.R.B. 749, 751, 1989 WL 223929, at *4-*5 (1989). On the other hand, recognition of a single unit threatens qualitatively greater disruption. The history of SIUH bears this out. A 1987 strike at the south site proved that separate units at separate sites can actually have a salutary effect. The north site remained open and cushioned the impact of the south site strike. Patients were transferred from the south site to other hospitals — including the north site — and north site employees filled in at the south site. The work stoppage threat is not compelling. The record also shows no significant whipsawing problem, or undue hindrance of RN transfers. In fact, RNs and their unions at both sites oppose involuntary intersite transfers.
SIUH asserts that the NLRB Regional Director wrongly applied the “overwhelming community of interests” standard to the unit determination issue, and that application of the “substantial community of interests” standard would lead to recognition of a single, institution-wide bargaining unit. SIUH reads too much into the Regional Director’s unit determination decision. He simply avoided repeating the entire factual analysis undertaken for the accretion issue. That analysis was relevant to the unit determination decision, and supported it.
Finally, SIUH asserts that the NLRB wrongly refused to consider new evidence. We disagree. SIUH sought to introduce evidence that it is the only metropolitan area acute care facility with separate RN bargaining units, and that separate units threaten to impede expansion and reorganization plans. The NLRB refused to hold a new hearing because the evidence post-dated the representation hearing and therefore was not admissible on reconsideration, and because the evidence in any event would not require a different result. We agree.
Petition for review denied; cross-petition for enforcement granted.
. The detoxification unit was moved in 1989 from the north site to the south site, and SIUH, NYSNA and then-south site RN representative Richmond Comity Nurses Association agreed to permit the RNs in that unit to remain in NYSNA despite their location at the south site.