To the Editor: A potentially serious problem with the performance of the Kallestad alpha-fetoprotein kit has been reported by Knight et al. (Feb. 20 issue).1 We tested 3563 second-trimester serum samples for alpha-fetoprotein, using the Kallestad kit. Median values for each gestational week were established in our laboratory, and results on routine quality-control procedures for the assay were consistently within normal limits. We found that the distribution of values was strongly skewed to the low end, with 17.5 percent falling below 0.5 multiples of the median and 11.1 percent falling below 0.4 multiples of the median. Our results were in.