This paper presents a review of OECD forecasts for the G7 countries for the period 1969 - 1997. The analysis deals with the magnitude, auto-correlation and eventual bias of the errors. In addition, we scrutinize the cyclical behavior of errors and try to identify the relationships between forecast values. In particular, we examine whether the Phillips curve relationship is the same for the actual data and the forecast values. The analyses clearly show that the forecasts are overly optimistic in terms of both output growth, public expenditure and inflation. Thus, for instance, output growth errors increase when output decreases and vice versa. The relationships between actual values and forecast values seem to differ but in a quite unsystematic way.