Amminger et al raise some interesting issues. I certainly agree that the estimation of premorbid IQ, particularly in patients with schizophrenia, is challenging and that further validation studies on methods for making such estimates should be pursued.

More specifically with reference to our earlier paper on the relationship of DUP to cognitive functioning (Normanet al, 2001), Amminger et al argue for the likely superiority of Bilder et al's (1992) index as a measure of cognitive deterioration in contrast to estimates based on NART-estimated premorbid IQ minus current WAIS full-scale IQ. In this respect they note that 38.1% of patients in a recent study by their group showed higher current IQ than NART-estimated premorbid IQ. This would, of course, suggest an increase in IQ after illness onset — an unlikely occurrence. I have examined this issue in our data-set and found such a pattern in 17.8% of our sample, with the average discrepancy among these individuals being 8.4 points. I can also confirm that in our sample, as in Amminger et al's sample, NART scores were correlated with age at admission (r=0.24,P<0.05), but WAIS—R full-scale scores were not.

The substantive question, of course, is whether DUP is related to cognitive deterioration. Amminger et al report that they have found DUP related to deterioration based on Bilder's index. We had reported some results using Bilder's index in our earlier paper. I will take this opportunity to report further that when we examined correlations between our two indices of DUP and Bilder's deterioration index they were non-significant (r=0.06 andr=0.04). We are currently pursuing the issue of whether DUP may be related to recovery of cognitive functioning during the first year of treatment.

The discrepancy between our earlier findings and those of Dr Amminger and colleagues does not appear to be explained on the basis of use of the NART rather than Bilder index. Other variables related to sample composition may be relevant. Also of potential importance is the method of measuring DUP, which, as has been suggested elsewhere (Norman & Malla, 2001), also needs to be more carefully considered and standardised. In this, as in many areas of psychiatric research, cumulative progress is dependent on careful and comparable measurement across studies. I endorse Amminger et al's comments in this respect.