This investigation was undertaken to determine how different the weather forecasts of telecasters are from those of the National Weather Service for the same areas and times, and the sources of information telecasters use when they modify the NWS forecast or develop their own. Questionnaires were mailed to 453 television weatherpersons, most of whom were seal holders of the AMS; 67 percent were returned. It was determined that 1) only 6% of all respondents do not consider the NWS forecast at all, whereas for almost 60% this consideration is moderate or heavy; 2) the percentages of those who transmit a.forecast different from that of the NWS 0–10%, 40–60% and 90–100% of the time are 16, 30 and 8, respectively; 3) when the forecast issued was different from that of the NWS, the elements most likely to be different were precipitation (84%), temperature (82), state of sky (66), and wind speed and direction (34); 4) the factors that influence the telecaster to issue a different forecast are, in order, current weather, local conditions, the telecaster's own analysis of forecast models, the telecaster's experience, satellite photographs, radar, and colleagues. The responses to each of these four items were further stratified by whether or not the respondent held the AMS Seal of Approval, whether he/she had a bachelor's degree, years as a telecaster, and the size of the market area. The division of responses by the first two of these stratifiers produced the largest changes from the percentages for all respondents given in 1 and 2 above and also produced relatively large differences within that stratifier. For example, forecasters who are least likely to consider the NWS forecast, and who issue a forecast different from that of the NWS a large percentage of the time, are seal holders with a degree and 6 to 10 yr experience who work in one of the top twenty market areas.