We investigated the longitudinal associations between physical activity (PA), lumbar multifidus morphology, and impactful low back pain (LBP) in young people. Nine-year-old children were recruited from 25 primary schools and followed up at age 13, 16, and 21 years. We measured PA with accelerometers at age 9, 13, and 16; quantified patterns of lumbar multifidus intramuscular adipose tissue (IMAT) change from 13 to 16 years using magnetic resonance imaging; and recorded LBP and its impact with standardised questionnaires and interviews. Associations were examined with crude and adjusted logistic or multinomial models and reported with odds ratios (OR) or relative risk ratios (RRR). We included data from 364 children (mean[SD] age = 9.7[.4] years). PA behaviour was not associated with LBP. Having persistently high IMAT levels at age 13 and 16 was associated with greater odds of LBP (OR[95% CI] = 2.98[1.17 to 7.58]). Increased time in moderate and vigorous intensity PA was associated with a lower risk of higher IMAT patterns (RRR[95% CI] = .67[.46 to .96] to .74[.55 to 1.00]). All associations became non-significant after adjusting for sex and body mass index (BMI). Future studies investigating the relationships between PA behaviour, lumbar multifidus IMAT, and impactful LBP should account for potential confounding by sex and BMI.