A Doppler velocity log measures relative velocity between an instrument and the bottom of a body of water by transmitting acoustic pulses that are scattered off the bottom. The scattered sound is received and the Doppler shift is measured. Like any other sensor, the data quality of a Doppler velocity log can be quantified by a variance and a bias. The goal of this work is to model error sources that do not average out over time, quantify those sources under various operating and environmental conditions, and derive a statistically-significant range of expected long-term errors. The following error sources are analyzed: absorption bias, terrain bias, side-lobe coupling, beam alignment, clock drift, and speed-of-sound error, which is a function of temperature, salinity, and pressure. Terrain bias is dependent on aperture diameter, aperture type (piston or phased-array), Janus angle, and bottom type. Absorption bias is dependent on aperture diameter, aperture type (piston or phased-array), Janus angle, the acoustic absorption coefficient, and altitude. Side-lobe coupling error is dependent on beam pattern, bottom slope, and instrument attitude. The magnitude of the remaining error sources depend on component tolerances and auxiliary sensor accuracies. The effect on the horizontal velocity of each error is analyzed in detail and the analysis is carried out under various environmental conditions and operating conditions.