We investigate the application of various correlation methods to the p
roblem of determining the temporal lag between flux maxima in light cu
rves produced by blazars in a flaring state. We generate simulated bla
zar light curves, and calculate the lag between different frequencies
using three different correlation methods. A Monte Carlo process allow
s estimates of confidence intervals of derived lags to be obtained, an
d these are shown to be large. Irregular sampling is shown to be a maj
or factor in increasing the width of the confidence intervals. In a co
mparison between the three cross-correlation methods, the linear inter
polation technique of Gaskell and Peterson (1987) performs better than
the discrete correlation function (DCF) of Edelson and Krolik (1988),
having narrower confidence intervals. It can, however, underestimate
the true lag thus lowering the chance of a significant detection. Esti
mating the lags by eye is seen to be surprisingly robust, and can be u
sed as a check on the other two techniques.