This paper analyzes interference and timeliness in the revenue-forecasting process, using new data on revenue-forecasting practices in low-income countries. Interference is defined as the occurrence of a significant deviation from purely technical forecasts. A theoretical model explains forecasting interference through government corruption. The data broadly supports the model, and the results are robust to alternative explanations. The paper also constructs three indices-transparency, formality, and organizational simplicity-that characterize revenue-forecasting practices, and assesses their effectiveness in producing an upfront-that is, timely-budget envelope. More transparent and simple forecasting processes lead to early budget constraints, while formality has no measurable effect.