This study had two main goals. First, an attempt was made to construct and validate an indicator of research performance through collecting peer judgments on the quality of journals by means of a world‐wide mail survey among 385 scholars. Second, to study the validity of indicators based on citations, these judgments were used to probe the quality of the coverage by the SSCI and the A & HCI of both core and noncore journals. Four disciplines in the humanities (General Linguistics, General Literature, Dutch Literature, and Dutch Language) and two disciplines in the social and behavioral sciences (Experimental Psychology and Public Administration) were studied. Coverage in both SSCI and A & HCI was generally increased somewhat when journals judged to be nonscholarly were eliminated. For non‐locally oriented disciplines, coverage of core journals was good: 85–100%. However, for locally oriented disciplines this varied between 20% and 40%. Despite limitations, the Journal Packet Quality indicator seems useful as a first, but crude approximation of the level of research performance when the number of articles is not too small. On an aggregate level, results showed convergence with those based on journal impact factors. © 1991 John Wiley & Sons, Inc.