Cyclic variability (CV) in combustion in IC-SI engines, manifested as variation in combustion pressure, is a long term as well as a contemporary problem. CV and its relation to several areas of engine control and diagnostics are discussed. A novel stochastic model relating combustion pressure to crankshaft velocity in an IC-SI engine is presented. This model incorporates a random combustion pressure signal model composed of a deterministic waveform and a random sequence modeling cyclic variation in combustion pressure. A discrete recursion is developed relating this sequence to noisy measurements of velocity. The inverse problem of reconstructing this sequence from one measurement of velocity per combustion is solved using a signal processing deconvolution method. Experimental results using real-world engine data are presented verifying the theoretical developments for low to moderate engine speed and moderate engine load. This reconstruction of cyclic pressure variability is simple to implement and may provide practical utility for engine control and diagnostics.