I estimate the local effects of gold mining on human capital accumulation in contexts where small-scale and illegal mining is prevalent. Evidence is based on high-resolution imagery of gold mining and rich administrative data from Colombia. I exploit the exogeneity of international prices and gold deposits to identify causal effects. Results indicate that mining increases enrollment in primary school and reduces dropout rates throughout the school cycle. However, it also reduces standardized test scores and college enrollment, particularly in academic degrees and STEM fields. I explore different mechanisms that could explain these results. Child labor is overall unaffected, but young adults between 19 and 25 are more likely to work in the mining sector. While municipal revenue and expenditure increases in mining municipalities, there are no substantial improvements in education spending or school inputs. Mining also intensifies violence, with potentially adverse effects on education.