Estimates of additive genetic, reciprocal and average heterosis effects were obtained from a three by three complete diallelcross experiment for giant freshwater prawn Macrobrachium rosenbergii. The data were generated in two batches. Three populations were sampled from the states of Gujarat (West), Kerala (South West) and Orissa (East) in India. The two mating batches of diallel crosses were combined in the analysis. A total 4773 animals (2233 in Batch 1 and 2540 in Batch 2) were tagged with VIA (Visible Implant Alpha numeric tags). A total of 2545 animals (979 and 1566 from Batch 1 and 2, respectively) were harvested with readable VIA tags and were; therefore available for analysis. The fixed effects included in the models were test environment (ponds 1 or 2 in batch 1, ponds 3 or 4 in batch 2), sex (male or female) and batch (1 or 2). Age at harvest was fitted as linear covariate. The genotype effect was partitioned into additive, non additive and reciprocal cross effects. Body weight, standard length and carapace length were recorded during tagging, sampling and harvesting. Estimates of additive genetic effect in the three stocks showed that the Kerala stock had greater body weight, carapace length and standard length, followed by Orissa then Gujarat. Average heterosis for body traits was low (3.1 to 6.3%) and not significantly different from zero. The results are discussed with respect to the choice of the most appropriate breeding strategy to improve the performance of M. rosenbergii in India.