Objectives To examine predictors of partial and full measles, mumps and rubella (MMR) vaccination catch-up between 3 and 5 years.

Design Secondary data analysis of the nationally representative Millennium Cohort Study (MCS).

Setting Children born in the UK, 2000–2002.

Participants 751 MCS children who were unimmunised against MMR at age 3, with immunisation information at age 5.

Main outcome measures Catch-up status: unimmunised (received no MMR), partial catch-up (received one MMR) or full catch-up (received two MMRs).

Results At age 5, 60.3% (n=440) children remained unvaccinated, 16.1% (n=127) had partially and 23.6% (n=184) had fully caught-up. Children from families who did not speak English at home were five times as likely to partially catch-up than children living in homes where only English was spoken (risk ratio 4.68 (95% CI 3.63 to 6.03)). Full catch-up was also significantly more likely in those did not speak English at home (adjusted risk ratio 1.90 (1.08 to 3.32)). In addition, those from Pakistan/Bangladesh (2.40 (1.38 to 4.18)) or ‘other’ ethnicities (such as Chinese) (1.88 (1.08 to 3.29)) were more likely to fully catch-up than White British. Those living in socially rented (1.86 (1.34 to 2.56)) or ‘Other’ (2.52 (1.23 to 5.18)) accommodations were more likely to fully catch-up than home owners, and families were more likely to catch-up if they lived outside London (1.95 (1.32 to 2.89)). Full catch-up was less likely if parents reported medical reasons (0.43 (0.25 to 0.74)), a conscious decision (0.33 (0.23 to 0.48)), or ‘other’ reasons (0.46 (0.29 to 0.73)) for not immunising at age 3 (compared with ‘practical’ reasons).

Conclusions Parents who partially or fully catch-up with MMR experience practical barriers and tend to come from disadvantaged or ethnic minority groups. Families who continue to reject MMR tend to have more advantaged backgrounds and make a conscious decision to not immunise early on. Health professionals should consider these findings in light of the characteristics of their local populations.