A simple method is proposed for predicting the fate of an epidemic outburst from early data. The method is based on the Richards model, and linearizations are proposed for obtaining preliminary values. A second step with nonlinear estimation fed with preliminary values as initial guess values may be attempted if field conditions allow the computation. The method was tested on data from 2001 dengue outbursts in both Havana and Winward Islands (French Polynesia). Predictions were satisfactory and an attempt of true prediction based on daily data for the 2009 H1N1 influenza outburst in the USA was undertaken. Comparison of early predictions with actual values obtained 3 months later suggests that some of the discrepancies are not due to method's inaccuracy, but to real improvement of infection rate as the H1N1 outburst proceeded. The method can be applied in any setting where cumulative number of cases is properly recorded.