The lack of empirical support for the positive economic impact of information technology (IT) has been called the IT productivity paradox. Even though output measurement problems have often been held responsible for the paradox, we conjecture that modeling limitations in production-economics-based studies and input measurement also might have contributed to the paucity of systematic evidence regarding the impact of IT, We take the position that output measurement is slightly less problematic in manufacturing than in the service sector and that there is sound a priori rationale to expect substantial productivity gains from IT investments in manufacturing and production management. We revisit the IT productivity paradox to highlight some potential limitations of earlier research and obtain empirical support for these conjectures. We apply a theoretical framework involving explicit modeling of a strategic business unit's (SBU)(1) input choices to a secondary data set in the manufacturing sector. A widely cited study by Loveman (1994) with the same dataset showed that the marginal contribution of IT to productivity was negative. However, our analysis reveals a significant positive impact of IT investment on SBU output. We show that Loveman's negative results can be attributed to the deflator used for the IT capital. Further, modeling issues such as a firm's choice of inputs like IT, non-IT, and labor lead to major differences in the IT productivity estimates. The question as to whether firms actually achieved economic benefits from IT investments in the past decade has been raised in the literature, and our results provide evidence of sizable productivity gains by large successful corporations in the manufacturing sector during the same time period.