The brain is a plastic organ, and so intraspecific studies that compare results obtained from wild individuals with those from common-garden experiments are crucial for studies aiming to understand brain evolution. We compared volumes of brain regions between reproductively isolated populations of a neotropical fish, Poecilia mexicana, that has locally adapted to perpetual darkness (Cueva Luna Azufre), toxic hydrogen sulphide in a surface stream (El Azufre) or a combination of both stressors (Cueva del Azufre). Wild fish showed habitat-dependent differences: enlarged telencephalic lobes and reduced optic tecta were found in fish living in darkness and sulphidic waters, in darkness without hydrogen sulphide or exposed to light and sulphide; fish from the sulphidic cave additionally showed enlarged cerebella. Comparison with common-garden reared fish detected a general decrease in brain size throughout populations in the lab, and little of the brain size divergence between lab-reared ecotypes that was seen in wild-caught fish. The pronounced differences in brain region volumes between ecotypes in the wild might be interpreted within the framework of mosaic evolution; however, the outcomes of common-garden experiments indicate a high amount of phenotypic plasticity. Our study thus highlights the importance of combining the investigation of brain size in wild populations with common-garden experiments for answering questions of brain evolution.