There is growing evidence that both sedentary behaviour and loneliness have adverse health consequences; however, the potential relationships between sedentary behaviour and loneliness are unclear. Accordingly, we examined the relationship and the directionality of prospective relationships between television (TV) viewing time (a ubiquitous leisure-time sedentary behaviour) and loneliness in middle-aged and older adults. We used data from three waves (2008-2013; follow-up every two years) of the English Longitudinal Study of Ageing cohort, in which middle-aged and older adults (3722 women and 3123 men) aged 50 years and older were included. The duration of TV viewing was quantified with two self-report questions, based on the hours spent on this behaviour on weekdays and weekends. Loneliness was assessed via the UCLA 3-item Loneliness Scale. Gender-stratified random-intercept cross-lagged panel models were constructed to examine the potential relationships between TV viewing and loneliness. Among middle-aged and older adults, loneliness in the previous assessment wave was significantly correlated with the level of loneliness in the following wave (women: β = 0.182, 95 % CI [0.096-0.267], men: β = 0.206, 95 % CI [0.116-0.297]). Moreover, there was evidence for a temporal directionality among women, but not men. That is, higher levels of loneliness in women predicted longer TV viewing duration in the subsequent wave (β = 0.159, 95 % CI [0.011-0.308]). This study identified gender-specific longitudinal relationships between TV viewing time and loneliness in middle-aged and older adults. We observed that loneliness can be a predictor of TV viewing time for women, might leading to increased screen-based leisure-time sedentary behaviour, which has been previously identified as an emerging risk factor for adverse events in specific health domains.