This article is concerned with the estimation of production functions, returns to scale, and measurement of the rate of technical change using panel data. Technical change is represented by single as well as multiple time trends. The underlying production technology is represented in translog functional form. A random effects model with heteroscedastic variances is used. The models are estimated using the generalized least squares method. The disturbances of cross-sectional units are assumed to be correlated over time. Empirically, our focus is on measuring technical change in Iranian manufacturing industries during the period 1971-1993. Empirical results show that single or multiple time trend representations yield different time behavior of technical change. In the multiple time trends model, we observe a sharp decline in the pattern of technical change in 1978 in relation to the political changes. In the single time trend, as expected, the sharp decline cannot be revealed due to the smooth pattern of technical progress during the entire period of study.