Few laboratories determine their normal values, because of the work involved. This study attempted to see whether a group of laboratories, each contributing a portion of the samples, could pool their information to yield data applicable to the whole group. Each of 20 VA Hospital laboratories obtained approximately 20 fasting early morning specimens from healthy men, aged 20-29 years, during a four-week period. Routine chemistry examinations (16) were performed on Technicon AutoAnalyzers (12/60, 6/60, and AAII), and model S Coulter Counters were used for five hematology tests (381 values for each constituent). Within-day, day-to-day, and interlaboratory variance were determined in parallel-method variability studies. The data showed: (1) it is valid to pool normal values from different laboratories when non-random variables have been eliminated; (2) non-random variables can be identified by increased interlaboratory variance, whether differences such as methods and equipment are obvious or not; (3) normal values so derived agree well with those from large series performed in single laboratories.