A signal analysis method used for condensing refrigerants is presented, which resulted in a modification of the prediction method for heat transfer based on probabilistic time-fractional results. An objective visual method for discrimination of flow patterns and determining probabilistic time-fractions in intermittent flow was developed. The frequency domain was identified as the main candidate for discrimination of sub-regimes present in intermittent flow. Experimental work was conducted using refrigerants R-22 and R-134a, at an average saturation temperature of 40°C, with mass fluxes ranging from 200–700 kg/m2s, and with test section inlet vapour qualities ranging from 0.65 down to 0.10. These test conditions mostly represented intermittent flow. The modified correlation predicted the experimental data with a mean absolute deviation of 10%.