The study of PASS 2 (program analysis of service systems) had two aims: to investigate the internal statistical properties of the evaluation instrument and to compare the relative quality (as measured by PASS 2) of services to the mentally retarded delivered in community versus institutional settings, to different age groups, in several human service fields, on the single service versus service system level, and in Canada versus the United States. The 102 service system evaluations comprising the nonrandom, accidental sample were those that were available to the investigator. Data analysis included the computation of basic descriptive statistics, correlation matrixes, internal consistency reliability coefficients, factor analysis, and stepwise regression. Major findings included that mean service quality was only modestly above the minimally acceptable level; the more important, serious, and universal the aspect of service quality, the less well was it being attained in practice; and an eight-item short form of PASS 2 was derived by means of stepwise regression. Results tended to support the prior conceptualization underlying both PASS 2 and the forthcoming PASS 3 without major exception. Several concluding suggestions were made concerning outstanding PASS research needs. (Tables with statistical data as well as a sample PASS checklist and score sheet are provided.) (Author/SBH)