Using x to predict y when they are highly correlated I am trying to use an organism's length (x) to predict weight (y), but length and weight are highly correlated (r=0.95). Are there any pitfalls I should be aware of because estimates of the two variables are in a sense measuring the "same" thing, i.e., size? I would be happy for someone to point out that this question has been asked before, but I don't see it on this site.
 A: You have apparently found a strong positive linear relationship. Assuming you're using some linear model for your prediction, this is excellent news(congratulations). 
It's quite intuitive that length would be strongly correlated with weight. Body volume times average body density would be even more so (and even more closely predictive-- exactly so). 
As with all predictions, the pitfalls you face are many and varied, depending on your specific model and purposes.
Make sure you didn't sample snakes to predict the weight of elephants (metaphorically or literally).
A: Perfect correlation does not mean a "1 to 1" relationship in the sense you are thinking.
Suppose I mapped X and Y with the line Y = B1 * X + B2 where B1 and B2 are coefficients that I find with a regression.  
If B1 = 1 and B2 = 0, and X and Y are perfectly correlated, then X and Y are the same thing.
Alternatively, if X and Y are perfectly correlated and my regression gives me B1 $\neq$ 1 and/or B2 $\neq$ 0, then X and Y are not the same thing.
