This study compared the white blood cell (WBC) and red blood cell (RBC) counts obtained with the Cell-Dyn 3200 (CD 3200) with results obtained by hemocytometer, the reference method for counting cerebrospinal fluid (CSF) and other body fluid specimens. Ninety-six CSF and 65 body fluid specimens were evaluated. Background counts were maintained on the CD 3200 at 0.001 x 10(9)/L and 0.00 x 10(12)/L for WBC and RBC counts, respectively. Linearity and precision were acceptable for both the total nucleated cell (TNC) count and the RBC count. The CD 3200 WBC optical count was correlated with the TNC count obtained by the manual reference method for CSF specimens across the range of 0 x 10(9 )/L to 7.863 x 10(9)/L (r2 = 0.9867) and for body fluid specimens across the range of 0 x 10(9)/L to 14.0 x 10(9)/L (r2 = 0.9955). An r2 value of 0.9016 was obtained for the 82 CSF specimens with manual TNC counts of <0.200 x 10(9)/L. Analysis of the CSF and body fluid specimens indicated that automated RBC counts could be reported at > or = 0.003 x 10(12)/L. In this study, 7 CSF and 30 body fluid specimens had RBC counts of >0.003 x 10(12)/L, and there was good agreement with manual RBC counts, with r2 values of 0.9893 and 0.9960 obtained for CSF and other body fluids, respectively. The CD 3200 in our experience has a lower reportable range than the ranges of most automated cell counters reported in the literature. In contrast to the only other instrument with comparable reportable ranges, the CD 3200 requires a smaller sample volume without any special sample preparation, reagents, or software. By using the CD 3200 with our laboratory-specific rules for agreement between duplicate counts, we would be able to reduce our manual CSF specimen counts from 192 TNC and 192 RBC counts to 2 TNC and 178 RBC counts. For body fluid specimens, our manual counts would be reduced from 130 TNC and 130 RBC counts to 10 TNC and 4 RBC counts.