Objective There is conflicting evidence on the association between television viewing in childhood and cognition. No recent UK and few worldwide studies have looked at the longer-term effects of television viewing in childhood. This study investigates the association between verbal and non-verbal cognitive outcomes at 7 years and television habits at 3, 5 and 7 years.

Method The study comprises longitudinal and cross-sectional analysis of the United Kingdom Millennium Cohort Study based on 8,148 children with complete data on variables of interest, maternal reports of television viewing and scores from objective tests of cognitive ability (British Ability Scales Second Edition) collected when cohort members were 3, 5 and 7 years. Mean ability scores were converted into the equivalent progress expected in a child at each age group using existing age equivalents derived for the cohort study population. Linear regression was used to estimate the relationship between each subscale and categories of television viewing (in relation to a reference group who watched between 1–3 hours of television a weekday) after adjustment for co-variates.

Results Children who did not watch television at 3 years had verbal ability scores at 7 years approximately 7 months behind the reference group (p < 0.05); their non-verbal ability skills were approximately 18months delayed (p < 0.05). Children who watched less than 1hour a weekday had delayed non-verbal skills of approximately 2 months (p < 0.05).

There was no significant association between television viewing at 5 and verbal or non-verbal ability at 7 years.

Children who did not watch television at 7 years had verbal ability scores approximately 3 months ahead of the reference group (p < 0.05). Those who watched less than 1hour a weekday were approximately 1month ahead (p < 0.05). Children who watched over 3 hours of television a weekday had delays of more than 1month (p < 0.05). There was no significant association between television viewing at 7years and non-verbal ability.

Conclusion There was a positive association between television viewing at 3 years and verbal and non-verbal cognition at 7 years in this cohort but a negative association between television viewing at 7 years and verbal skills at 7 years. This may influence potential guidance on television viewing. Possible explanations are the type of programmes watched or accumulation of viewing hours.