We investigate the external validity of corruption experiments by conducting the same experiment in three different environments: a laboratory in a developed country, a laboratory in a developing country and the field in a developing country. In the experiment, a candidate proposes a bribe to a grader to obtain a better grade. We find the direction and magnitude of several treatment effects to be statistically indistinguishable across the three environments. In particular, increasing the graders’ wage reduces the probability of accepting the bribe but promotes reciprocation. Our results therefore provide evidence that laboratory experiments on corruption can have empirical relevance.