Processing modifications were made to the COBAS AMPLICOR HCV version 2.0 assay to enhance sensitivity. Two methods of specimen concentration, centrifugation ("ultraspin") and cationic detergent plus silica membrane ("ultracolumn"), were compared to the standard method. The effect of these changes on assay sensitivity and specificity was examined using commercial hepatitis C virus (HCV) preparations. The limits of detection (LOD, defined as detection of HCV RNA in >/= 95% of replicates) of genotype 1a were 50, 12, and 6 by standard method, ultraspin and ultracolumn, respectively. For genotype 1b, the LOD was 25 IU/ml, 12 IU/ml, and 3 IU/ml; for 2b, it was 50, 12, and 3; for 3a, it was 25, 12, and 1.5; for 4 it was 18, 4, and 2; for 5a, it was 38, 9, and 2; and for 6a it was 47, 6, and 3. No false positives were detected after ultraspin when controls containing high or low HCV concentrations were alternated with normal human plasma. Plasmas in which HCV RNA was not detected by the standard assay were re-tested with modified methods to assess the effect of altered processing in clinical specimens. Three of 152 specimens with no detectable HCV RNA by the standard method were positive by ultraspin and 2 of 109 were positive by ultracolumn, suggesting that these methods may increase assay sensitivity in clinical specimens.