The Ultraviolet/Optical Telescope (UVOT) on board Swift has the capability to provide critical insight into the physics of the early afterglows of gamma-ray bursts (GRBs). But without precise calibration of the UVOT to standard photometric systems, it is impossible to leverage late-time, ground-based follow-up data to the early-time UVOT observations. In this paper, we present a calibration of the Swift UVOT photometry to the standard Johnson UBV system for the UVOT U, B, and V filters, and a step-by-step photometry recipe for analyzing these data. We base our analysis on aperture photometry performed on the ground-based and UVOT observations of the local standard stars in the fields of supernovae (SNe) 2005am and 2005cf and a number of Landolt standard stars. We find that the optimal photometry aperture radius for UVOT data is small (2.5`` for unbinned data, 3.0’’ for 2×2 binned data), and we show that the coincidence-loss (C-loss) correction is important even for relatively faint magnitudes (16-19 mag). Based on a theoretically motivated model, we fit the C-loss correction with two parameters, the photometric zero point (ZP) and the saturation magnitude (m$_ınfty$), and derive tight constraints for both parameters [σ(ZP)=0.01 mag and σ(m$_ınfty$)=0.02 mag)]. We find that the color-term correction is not necessary for the UVOT B and V filters but that it is necessary for the U filter for blue objects [(U-V)<0.4 mag]. We analyze the UVOT UBV photometry of SN 2005am and find that the UVOT photometry is generally consistent with the ground-based observations but that a difference of up to 0.5 mag is found when the SN became faint. We also apply our calibration results to the UVOT observations of GRB 050603. There is a scatter of åisebox-0.5ex 0.04-0.08 mag in our final UVOT photometry, the cause of which is unclear, but it may be partly due to the spatial variation in the pixel sensitivity of the UVOT detector.