This paper considers the stochastic production frontier approach to analyse the technical efficiency and its determinants of fish farms in Ghana using a cross-section data of 150 farms. It considers the explicit effects of family and hired labour on production by setting the log-value of the zero-observation of these two sources of labour to zero with dummy variables. Results demonstrate that expected elasticities of mean output with respect to all input variables are positive and significant. Findings also show that family and hired labour used for fish farming in Ghana may be equally productive. Fish farms in Ghana are revealed to be characterised by technology with increasing return to scale. The combined effects of operational and farm specific factors are found to influence efficiency. The study further reveals that inclusion of interaction between some exogenous variables in the inefficiency model is significant in explaining the variation in efficiency. Results also suggest that small pond operators are more efficient than farms with large ponds. Mean technical efficiency is estimated to be 78 percent.