I need to combine two measurements (morning and evening Peak Flow Rate) in a single daily score. I don't have individual values, just the means and standard deviations. How should I compute the estimated SD of their average? My guess was that the SD of the sum of two means should be the same as that of their difference, the pooled SD, for which the formula is easily found in textbooks. (Since SD(a-b)= SD(a-(-1*b), where a and b are two samples). But then, when I divide the sum by two to obtain the average, I must divide also the SD (since SD(a/2)=SD(a)/2), and the resulting SD appears to be unrealistically low. Where is the catch? Thanks, Piersante Sestini Institute of Respiratory Diseases University of Siena Italy %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%