OBJECTIVES: This is an investigation into the longitudinal patterns of bed sharing, the characteristics associated with those patterns, and the relationship with breastfeeding. METHODS: The study used prospective, population-based data from the United Kingdom to investigate nocturnal bed sharing at 5 time points from birth to 4 years of age. Of 14 062 live births, 7447 (53%) had data available for all time points. RESULTS: Latent class analysis identified 4 mutually exclusive groups, broadly described as nonsharers (66%), early bed sharers (only in infancy) (13%), late bed sharers (after the first year) (15%), and constant bed sharers (throughout the 4 years) (6%). The boy/girl ratio and the proportion of families of nonwhite ethnicity were slightly higher in all 3 bed-sharing groups, compared with the non-bed-sharing group. Higher maternal educational achievement and higher social class were positively associated with early bed sharing, negatively associated with late bed sharing, and not associated with constant bed sharing. The 3 bed-sharing patterns were related significantly to breastfeeding at 12 months (P <.001), whether the families shared beds late (odds ratio: 1.72 [95% confidence interval: 1.36-2.18]), early (odds ratio: 2.36 [95% confidence interval: 1.87-2.97]), or for the whole period (odds ratio: 5.29 [95% confidence interval: 4.05-6.91]). The prevalence of breastfeeding was significantly higher among the groups that shared beds constantly or early for each of the first 15 months after birth. CONCLUSIONS: Advice on whether bed sharing should be discouraged needs to take into account the important relationship with breastfeeding.