The feeding behavior of the whitefly Bemisia tabaci (Gennadius) was monitored using the electrical penetration graph (EPG) technique during the transmission process of tomato yellow leaf curl geminivirus (TYLCV). The behavior of individual viruliferous whiteflies was recorded on two-leaf stage tomato test plants (Lycopersicon esculentum Mill ‘Riofuego’). A total of 213 whitefly individuals was recorded on single test plants during an inoculation access period that ranged from 3.5 to 14 h. Recordings were classified into 4 categories depending of the waveforms observed: group I, including only waveform C (pathway), was associated with a residual 2.4% TYLCV transmission efficiency; group II, showing stylet pathway and a single E(pd)₁ waveform, was associated with a 7.4% transmission efficiency; group III, showing stylet pathway and a single E(pd)₁ E(pd)₂ waveform, achieved a 23.4% transmission efficiency; and finally group IV, showing pathway followed by several E(pd)₁ E(pd)₂ waveforms, was associated with the highest transmission efficiency (37.5%). A total of 16 different behavioral variables was introduced into a stepwise-backward logistic regression model to determine the variables most related to TYLCV inoculation. Among them, the total duration of E(pd)₁ was the most significant variable associated with virus inoculation by B. tabaci (P = 0.002, positive relationship). In addition, the regression analysis indicated a strong positive relationship (P = 0.005) between transmission efficiency and total number of E(pd)₁ (t = 4.39, β = 0.45 ± 0.10, r = 0.87), and total duration of waveform E(pd)₁ (P = 0.05) (t = 4.00, β = 0.02 ± 0.004, r = 0.94). The minimum phloem contact E(pd)₁ threshold period observed for successful inoculation of TYLCV by B. tabaci on tomato plants was 1.8 min. Implications of these results for persistent virus transmission by whiteflies are discussed.