This paper estimates a sequence of energy-saving technology choices by Dutch glasshouse firms. The model allows for time-constant, firm-specific effects and serial correlation of errors and it is estimated on panel data over the period 1991¿1995. The unobserved error sequences are simulated in the model in such a way that they are consistent with the observed choices. The results suggest that positive revenue shocks encourage adoption of new technologies. The choices also exhibit firm-specific persistency over time. An increase of the price of energy encourages the adoption of a highly efficient energy-saving technology. The probability of investing in energy-saving technologies also decreases with capital invested in structures (e.g. double glazing of glasshouses) and increases with capital in energy installations.