More than 26 billion juveniles of 180 marine species are released annually into the wild in over 20 countries, but the usefulness of this strategy remains unclear. Here, I analyse the effects of stocking by Japanese marine and salmon stock-enhancement programmes and evaluate their efficacy through a Bayesian meta-analysis of new and previously considered cases. The posterior mean recapture rate (± SD) was 8.3 ± 4.7%. Without considering personnel costs and negative impacts on wild populations, the mean economic efficiency was 2.8 ± 6.1, with many cases having values of 1 to 2. On the macro-scale, the proportion of released seeds to total catch was 76 ± 20% for Japanese scallop, 28 ± 10% for abalone, 20 ± 5% for swimming crab, 13 ± 5% for kuruma prawn, 11 ± 4% for Japanese flounder, and 7 ± 2% for red sea bream; according to these percentages, stocking effects were generally small, and population dynamics were unaffected by releases but dependent on the carrying capacity of the nursery habitat. All cases of Japanese releases, except for Japanese scallop, were probably economically unprofitable. Captive breeding reduces the fitness of hatchery fish in the wild. In addition, long-term releases replace wild genes and may cause fitness decline in the recipient population when the proportion of hatchery fish is very high. Short-term hatchery stocking can be useful, particularly for conservation purposes, but large-scale programmes may harm the sustainability of populations. Nursery habitat recovery and fishing pressure reduction often outperform hatcheries in the long run.