The factors controlling lightning activity over central Israel and the adjacent Mediterranean Sea were studied. Potential predictors were correlated at 12-h intervals with total number of flashes. Since during the winter season lightning is generated in this region by Cyprus Lows, the data includes 283 observations on days in which this system prevailed for December to February, which covered four winters. The average lightning rate was 26.8 h-1, with a high standard deviation of 55.2 h-1. The total number of flashes at night exceeded the daytime number by 35%, in agreement with previous studies. The CAPE values were on the order of hundreds of J kg-1. A statistical linear multi-regression model was developed for the number of lightning flashes based on 35 atmospheric variables. The correlation between the modeled and the observed number of lightning flashes was 0.67, and 0.81 for the logarithm of the number of lightning flashes (log-lightning). This suggests that the lightning intensity responds exponentially to its governing factors. A linear multi-regression stepwise model for the log-lightning selected seven predictors as significant and yielded a correlation of 0.74. This model was validated by three holdout and three leave-one-out validation experiments. The composition and hierarchy of the significant predictors reflect the dominance of the thermodynamic factors, in particular instability, in determining lightning activity. Though thunderstorms are local or meso-scale phenomena, the synoptic-scale atmospheric variables were found to be powerful predictors for their intensity.