In order to measure the variations of the magnetic chemically peculiar star CU Vir, we obtained 2821 good Stroemgren uvby photometric values from the Four College Automated Photometric Telescope in the period 1998-2012 and 5529 values from the Solar Mass Ejection Imager in the period 2003-2006, allowing us to further clarify the situation concerning its rotational period, described previously by Pyper et al. (1998A&A...339..822P). Our main result is that the O-C data since 1993 are consistent with a constant period of 0.5207137+/-0.0000010d, which is longer than the period 0.52070308d of Pyper et al. (1998A&A...339..822P). The data during the time period 1993-2012 show no evidence of glitches. The data between 1987 and 1993 correspond to a shorter period of 0.520701d, and the inclusion of those data in the previous study made the published period come out shorter. We also cannot rule out a constantly lengthening period if the data since 1987 are included; however, the fit is slightly better for a constant period. At least five more years of consistent data will be required to distinguish between these two models. Radio astronomy data published since 1998 show phase shifts in the radio emission peaks that are inconsistent with the rotation period inferred from the optical photometry. These shifts may be due to instabilities in the region where the radio emissions originate; they indicate that the photosphere and the upper atmosphere are not rigidly coupled.