This paper introduces technological uncertainty into a timing game of new technology adoption. It is shown that the timing neither necessarily involves first-mover advantages in precommitment equilibria (Reinganum, Review of Economic Studies, XLVIII (1981) 395-405) nor rent-equalization due to the threat of preemption (Fudenberg and Tirole, Review of Economic Studies, LII (1985) 383-401). Rather, there may be second-mover advantages because of informational spillovers. Furthermore, the model predicts that the equilibrium payoffs will typically be discontinuous and non-monotonic in the probability that the new technology is profitable. A welfare analysis reveals several market failures, and suggests that policy intervention should adequately depend on the nature of uncertainty and the rate of technological progress. (C) 2000 Elsevier Science B.V. All rights reserved. JEL classification: L13; O31; O32.