Previous research has demonstrated adults' difficulties with explicitly forecasting exponential processes. Exponential growth is usually grossly underestimated, whereas exponential decline is forecast more accurately. By contrast, the present study examined implicit knowledge about exponential processes and how it is affected by function type (growth versus decline) in samples of 7-, 10-, 14-year-olds, and adults (N=80). Different indicators of the quality of forecasts were investigated. As opposed to previous findings, participants of all age groups estimated exponential decline less adequately than exponential growth. This effect could be attributed mainly to the fact that, in relation to fitted exponential functions, the starting value, or intercept, of the function was approximated well for exponential growth but badly with regard to exponential decline. The accuracy of the non-linear component in forecast functions barely differed between function types within the same age group. Furthermore, even 7-year-olds appeared to have a preliminary understanding of exponential processes, while both intercepts and exponents of forecasts became more accurate with age. Theoretical and practical implications are discussed.