Error in a regression

Hi Andrea,
Assuming that your model is something like:

lm(y~x,data=mydata)

See what:

cor(mydata$y,mydata$x)

returns. If it is very very close to 1 or -1, there lies your problem.
If one or more of your predictor variables is an almost perfect
predictor of the response, you don't have much room to select
different models.

Jim