Background: Motor dysfunction in the contralateral hand has been well characterised after stroke. The ipsilateral hand has received less attention, yet may provide valuable insights into the structure of the motor system and the nature of the recovery process. By tracking motor function of both hands beginning in the acute stroke period in patients with cortical versus subcortical lesions, we sought to understand the functional anatomy of the ipsilateral deficit. Methods: We examined 30 patients with first-ever unilateral hemiparetic stroke, 23 with subcortical lesions affecting the corticospinal tract, seven with cortical involvement. Patients performed hand dynamometry and the 9-Hole Peg Test (9HPT) with each hand at 24-48 h, 1 week, 3 months and 1 year after stroke. Linear regression was used to compare the two different motor tasks in each hand. Repeated measures ANOVA was used to compare recovery rates of the two tasks in the first 3 months. Results: Ipsilateral 9HPT scores averaged z = -7.1, -3.6, -2.5 and -2.3 at the four time points whereas grip strength was unaffected. The initial degree of impairment of grip strength in the contralateral hand did not correlate with the degree of impairment of 9HPT in either the contralateral or ipsilateral hand (r = 0.001, p = 0.98), whereas the initial degree of impairment of 9HPT in the contralateral hand correlated with the degree of impairment of 9HPT in the ipsilateral hand (r = 0.79, p = 0.035). The rate of recovery also differed for the two tasks (p = 0.005). Conclusion: Ipsilateral motor deficits are demonstrable immediately after stroke and extend into the subacute and chronic recovery period. Dissociation between grip strength and dexterity support the notion that dexterity and grip strength operate as anatomically and functionally distinct entities. Our findings in patients with subcortical lesions suggest that the model of white matter tract injury needs to be refined to reflect the influence of a subcortical lesion on bi-hemispheral cortical networks, rather than as a simple "severed cable" model of disruption of corticofugal fibres. Our data have implications for both stroke clinical trials and the development of new strategies for therapeutic intervention in stroke recovery.