This letter is in response to a paper entitled, "Hematologic Effects of Benzene: A Thirty-Five Year Longitudinal Study of Rubber Workers," published in Volume 4, No.4 in 1988. The data used in this paper was abstracted from records obtained under a Freedom of Information request from the National Institute for Occupational Safety and Health (NIOSH) in 1985. NIOSH investigators have also assembled this data and have been working to interpret it for several years. We are nearing completion of our own study, but feel it is necessary to comment on the findings presented in your journal at this time. Kipen et al. observed an increase in red cell and white cell counts during the period 1940 through 1948, and concluded that there was inverse relationship with exposure because WBCs in particular were negatively correlated with one of two available estimates of benzene dose (Crump and Allen, 1984). We observed the same temporal trend in blood cell counts during this period, but do not conclude that it is necessarily exposure related. We first examined temporal trends for routine blood samples (i.e., those not taken for pre-employment examinations or to repeat abnormal blood tests) by plotting the monthly mean blood counts over time. We observed that there was a marked shift upward occurring between July and August, 1947, in both RBC and WBC. Fitting two regression lines to the data before and after August 1, 1947, revealed that not only was there an upward shift between July and August, 1947, but there was also a significant trend in the data prior to the shift. Figure 1 illustrates this phenomenon for WBC during the period 1940- 1948 analyzed in the Kipen paper. The trend was similar for RBC. We hypothesized that the observed temporal changes might be due to something other than exposure, such as changes in laboratory procedures.