The Iowa Tests of Educational Development vocabulary test was administered once to ninth graders following standard procedures (ITED-V) and again using the alternate form in which students listened to a tape-recording of items (PROMPTED-V). Word meaning knowledge was controlled by entering the PROMPTED-V score first into a multiple regression equation to predict reading comprehension. The ITED-V score made a significant addition to the prediction equation, suggesting that the ability to decode printed words is a component of comprehension performance. Poor decoders were operationally defined as students with an ITED-V score more than two standard errors below their PROMPTED-V score. Their reading comprehension, as indicated by several measures, was significantly below that of good decoders who showed no discrepancy between their PROMPTED-V and ITED-V scores. Among poor decoders, there was a range in comprehension performance which suggests the possible influence of a third component of comprehension, such as ability to use contextual prediction.