This paper addresses two questions regarding models of service quality: single- vs. two-predictor models and items vs dimensions. This paper derives six alternative models specified by a two-way factorial design: 3 types of models (SERVQUAL vs. SERVPERF vs. two-predictor model) ×  2 abstraction levels (5 dimensions vs. 22 items). It compares the alternative models with the criteria of predictability, interpretability, and usefulness. It analytically examines the alternative models with the linkages between a theoretical three-predictor model and the alternative models, and empirically compares the alternative models based on data sets collected in five service industries with a convenient sampling technique. The analyses show that the two-predictor model is superior to SERVQUAL and SERVPERF in view of the three criteria. The two-predictor model allows one to examine the effects of expectation and performance on customer satisfaction. This property is important because the management of expectation is as important as the management of performance in managing customer satisfaction. It is possible to recover the two-predictor model as well as SERVPERF from a data set required for SERVQUAL. Thus, it is reasonable to use the two-predictor model as well as SERVQUAL and SERVPERF for gaining more meaningful insights on service quality.