Abstract Recovery of dexterous hand use is critical for functional outcome after stroke. Grip force recordings can inform on maximal motor output and modulatory and inhibitory cerebral functions, but how these actually contribute to recovery of dexterous hand use is unclear. This cohort study used serially assessed measures of hand kinetics to test the hypothesis that behavioural measures of motor modulation and inhibition explain dexterity recovery beyond that explained by measures of motor output alone. We also investigated the structural and functional connectivity correlates of grip force control recovery. Eighty-nine adults (median age = 54 years, 26% females) with first-ever ischaemic or haemorrhagic stroke and persistent arm and hand paresis were assessed longitudinally, at 3 weeks, and at 3 and 6 months after stroke. Kinetic measures included: maximal grip force, accuracy of precision and power grip force control, and ability to release force abruptly. Dexterous hand use was assessed clinically with the Box and Block Test and motor impairment with the upper extremity Fugl-Meyer Assessment. Structural and functional MRI was used to assess weighted corticospinal tract lesion load, voxel-based lesion symptom mapping and interhemispheric resting-state functional connectivity. Fifty-three per cent of patients had severe initial motor impairment and a majority still had residual force control impairments at 6 months. Force release at 3 weeks explained 11% additional variance of Box and Block Test outcome at 6 months, above that explained by initial scores (67%). Other kinetic measures did not explain additional variance of recovery. The predictive value of force release remained significant when controlling for corticospinal tract lesion load and clinical measures. Corticospinal tract lesion load correlated with recovery in grip force control measures. Lesions involving the parietal operculum, insular cortex, putamen and fronto-striatal tracts were also related to poorer force modulation and release. Lesions to fronto-striatal tracts explained an additional 5% of variance in force release beyond the 43% explained by corticospinal injury alone. Interhemispheric functional connectivity did not relate to force control recovery. We conclude that not only voluntary force generation but also force release (reflecting motor inhibition) are important for recovery of dexterous hand use after stroke. Although corticospinal injury is a main determinant of recovery, lesions to integrative somatosensory areas and fronto-parietal white matter (involved in motor inhibition) explain additional variance in post-stroke force release recovery. Our findings indicate that post-stroke upper limb motor impairment profiling, which is essential for targeted treatment, should consider both voluntary grasp generation and inhibition.