This study makes an attempt to examine the impact of errors in the specification and measurement of capital input on the estimates of time series production functions in Indian manufacturing. For this purpose, it examines the main alternative measures of capital input generally used in the production function exercises and also derives the corresponding set of alternative estimates of the widely used Cobb-Douglas production function for the Indian iron & steel industry, on the basis of time series data for the period 1946-1966, essentially in the form of an illustrative exercise. The major conclusions of the study are that the estimates of time series production functions for Indian manufacturing are highly sensitive to the measurement of capital input; and the elimination of preliminary errors in the measurement of capital input alters significantly the basic structure of the estimated production function and, hence, the estimates of the relative contributions made by various factors to the growth of output in Indian manufacturing.