Seasonal patterns in freezing tolerance of five Rhododendron cultivars that vary in feezing tolerance were estimated. Electrolyte leakage was used, and raw leakage data were transformed to percent leakage, percent injury, and percent adjusted injury. These data were compared with visual estimates of injury. Percent adjusted injury was highly correlated (0.753) to visual estimates. Two asymmetric sigmoid functions—Richards and Gompertz—were fitted to the seasonal percent adjusted injury data for all cultivars. Two quantitative measures of leaf freezing tolerance—Lt50 and Tmax (temperature at maximum rate of injury)—were estimated from the fitted sigmoidal curves. When compared to the General Linear Model, the Gompertz function had a better fit (lower mean error sum of squares) than Richards function. Correlation analysis of all freezing tolerance estimates made by Gompertz and Richards functions with visual LT50 revealed similar closeness (0.77 to 0.79). However, the Gompertz function and Tmax were selected as the criteria for comparing relative freezing tolerance among cultivars due to the better data fitting of Gompertz function (than Richards) and more descriptive physiological representation of Tmax (than LT50). Based on the Tmax (°C) values at maximum cold acclimation of respective cultivars, we ranked `Autumn Gold' and `Grumpy Yellow' in the relatively tender group, `Vulcan's Flame' in intermediate group, and `Chionoides' and `Roseum Elegans' in the hardy group. These relative rankings are consistent with midwinter bud hardiness values reported by nurseries.