Lightcurve observations of an asteroid provide strong constraints on its rotation state. Lightcurves are usually available from a much wider range of times and viewing geometries than other types of observations such as radar or occultations (which, for most asteroids, are not available at all). However, in some cases, lightcurves are reported with no uncertainties. In other cases, the reported error bars are clearly less than the typical scatter between neighboring data points, perhaps because the error bars do not account for possible systematic effects. When doing detailed modeling of an asteroid, in order to find its shape, rotation state, and other physical properties, one wishes to have error bars that reflect the true uncertainties of the data. Underestimated error bars may cause modeling software to introduce questionable features into the model, to match variations in a lightcurve that are below the level of the actual noise. A user can manually assign greater uncertainties to some lightcurves or to individual data points in cases where this seems necessary, but this is subjective; it is preferable to have a mathematical justification for making such changes. Following the work of Harris et al. 1989 (“Photoelectric Observations of Asteroids 3, 24, 60, 261, and 863”, Icarus 77, pp. 171-186) and others, based on fitting harmonics to a lightcurve, I present automated methods for adjusting lightcurves’ error bars (if necessary) and for detecting outliers. I also discuss the statistics of lightcurve residuals, which do not necessarily follow a Gaussian distribution.