Fast voltage-sensitive dyes, are widely used to image cardiac electrical activity. Typically, the emission spectrum of these fluorochromes is wavelength-shifted with altered membrane potential, but the optical signals obtained also decay with time and are affected by contraction. Ratiometry reduces, but may not fully remove these artifacts. An alternate approach has been developed, in which the time decay in simultaneously acquired short and long wavelength signals is characterized nonparametrically and removed. Motion artifact is then identified as the time-varying signal component common to both decay-corrected signals and subtracted. The performance of this subtraction technique was compared with ratiometry for intramural optical signals acquired with a fiber-optic probe in an isolated, Langendorff-perfused, pig heart preparation (n=4), stained with di-4-ANEPPS. The perfusate concentration of 2,3-BDM was adjusted (7.5 - 12.5mM) to alter contractile activity. Short (520-600 nm) and long (>600 nm) wavelength signals were recorded over 8-16 cardiac cycles at 6 sites across the LV free wall in sinus rhythm and during pacing. A total of 451 such data sets were acquired. Appreciable wall motion was observed in 225 cases, with motion artifact classed as moderate (< modulation due to the AP) in 187 and substantial (> modulation due to the AP) in 38. In all cases, subtraction performed as well as, or better than, ratiometry in removing motion artifact and decay. AP morphology was recovered more faithfully by subtraction than ratiometry in 58/187 and 31/38 cases with moderate and substantial motion artifact, respectively. This novel subtraction approach may therefore provide a means of reducing the concentration of uncoupling agents used in cardiac optical mapping studies.