We report the results of a worldwide campaign to observe WZ Sagittae during its 2001 superoutburst. After a 23-year slumber at V=15.5, the star rose within 2 days to a peak brightness of 8.2, and showed a main eruption lasting 25 days. The return to quiescence was punctuated by 12 small eruptions, of ~1 mag amplitude and 2 day recurrence time; these "echo outbursts" are of uncertain origin, but somewhat resemble the normal outbursts of dwarf novae. After 52 days, the star began a slow decline to quiescence. Periodic waves in the light curve closely followed the pattern seen in the 1978 superoutburst: a strong orbital signal dominated the first 12 days, followed by a powerful /common superhump/ at 0.05721(5) d, 0.92(8)% longer than P_orb. The latter endured for at least 90 days, although probably mutating into a "late" superhump with a slightly longer mean period [0.05736(5) d]. The superhump appeared to follow familiar rules for such phenomena in dwarf novae, with components given by linear combinations of two basic frequencies: the orbital frequency omega_o and an unseen low frequency Omega, believed to represent the accretion disk's apsidal precession. Long time series reveal an intricate fine structure, with ~20 incommensurate frequencies. Essentially all components occurred at a frequency n(omega_o)-m(Omega), with m=1, ..., n. But during its first week, the common superhump showed primary components at n (omega_o)-Omega, for n=1, 2, 3, 4, 5, 6, 7, 8, 9 (i.e., m=1 consistently); a month later, the dominant power shifted to components with m=n-1. This may arise from a shift in the disk's spiral-arm pattern, likely to be the underlying cause of superhumps. The great majority of frequency components ... . (etc., abstract continues)

PDF, 54 pages, 4 tables, 21 figures, 1 appendix; accepted, in press, to appear July 2002, PASP; more info at http://cba.phys.columbia.edu/