The power relationship between number of taxa and sampled area is one of the most cited and general patterns in ecology.  The relationship also forms the basis for some conservation and management programs.  However, several aspects of taxa-area relationships (TAR) remain unstudied.  TARs are typically assumed to be temporally invariant—an assumption that can have far-reaching consequences when made errantly.  I used data from a previous experiment on benthic macroinvertebrates to examine TARs in two streams.  The goal was to examine how TARs differed, both between streams and through time over a period that included floods and drought.  The experiment produced two surprising results.  1) z-value (the slope of the estimated TAR) of the two streams was very similar (0.346 & 0.348) to begin the experiment. However, z-values diverged through time, apparently because of differential drought effects.  Surprisingly, the z of the heavily impacted stream remained nearly constant through time, while z of the less-impacted stream changed significantly.  2)  In the stream with variable z, hydrologic condition explained over 70% of the variance, with z decreasing with increasing discharge.  TARs in small streams may vary considerably through time, but simple relationships with environmental variables may explain much of this variability.