The standard lymphocytotoxicity assay (LCT), a biotin-avidin enzyme immunoassay (ELISA), platelet suspension immunofluorescence test (PSIFT), and platelet radioactive antiglobulin test (PRAT) were examined in prospective crossmatching for selection of compatible random donor platelets for refractory patients. One hundred seven episodes of pooled random donor platelet transfusions were evaluated in 26 patients. There was good reproducibility of results by individual techniques. Concordance of results by the different methods was 40-60%. One-hour and 24 hr posttransfusion corrected count increments (CCI) were compared as parameters for assessing success or failure of the transfusion. Using a rank scoring system, the relative efficiency of predictiveness for all transfusions was PRAT greater than LCT greater than PSIFT greater than ELISA. Combination of PRAT and LCT afforded the best predictability and sensitivity was higher than for either PRAT or LCT alone (93 vs. 79 and 62%, respectively). Mean posttransfusion CCI (x 10(9)/L) following PRAT-compatible platelets was 13.9 +/- 12.7 at 1 hr and 7.3 +/- 6.9 at 24 hr; following PRAT-incompatible platelets, 5.7 +/- 7.8 (1 hr) and 2.1 +/- 4.1 (24 hr). Results were similar for LCT-tested platelets. A radioimmunofiltration modification of the PRAT developed and used in selected cases was simple, fast, efficient, and inexpensive. The study indicated that the techniques evaluated are practical and feasible for routine use in the provision of compatible random donor platelets to the refractory patient who has no other cause for increased platelet destruction.