This study aimed to investigate changes in non-cycloplegic ocular biometrics during the initial six months of treatment with a 0.1% atropine loading dose and 0.01% atropine compared with a placebo and analyze their contribution to the treatment effect on cycloplegic spherical equivalent (SE) progression. The study was based on a randomized, double-masked, placebo-controlled, multicenter trial evaluating a 0.1% atropine six-month loading dose and 0.01% atropine in reducing myopic progression in Danish children. The treatment phase was 24 months, and the washout phase was 12 months. Parameters measured included changes in axial length (AL), anterior chamber depth (ACD), lens thickness (LT), vitreous chamber depth (VCD), and choroidal thickness (ChT), while cycloplegic SE and lens power were calculated. Longitudinal changes and contributions to treatment effects were analyzed using constrained linear mixed models and mediation analyses, respectively. After six months, AL was 0.13 mm shorter (95% confidence interval [CI], −0.18 to −0.07 [adjusted p < 0.001]) and 0.06 mm shorter (95% CI, −0.11 to −0.01 [adjusted p = 0.060]) with a 0.1% atropine loading dose and 0.01% atropine, respectively, compared to the placebo group. Similar concentration-dependent changes were found with ACD, LT, VCD, ChT, and cycloplegic SE. Although the treatment effects trended toward concentration-dependent responses, only the treatment effect mediated by AL at three months differed significantly between 0.01% atropine and a 0.1% atropine loading dose (adjusted p = 0.023). Several ocular biometrics, including AL, ACD, and LT, changed dose-dependently during low-dose atropine treatment. Moreover, the treatment effect of atropine on SE progression was mediated by a subset of ocular biometrics, mainly AL, with trends toward concentration dependency and distributional shifts over time.