BACKGROUND Previous literature hypothesized that Winters type I are mainly characterized by a hypo-activation of dorsiflexors and type II by hyperactivation of plantarflexors around initial contact. However, it is currently not known if hemiplegic children belonging to the same Winters class really share the same muscle activation patterns, although this information might have relevant clinical implications in the patient management. METHODS Gait data of 38 hemiplegic cerebral palsy children (16 Winters type I, 22 Winters type II) were analyzed, focusing on the foot and shank. A 2.5-minute walk test was considered, corresponding to more than 100 gait cycles for each child, analyzing the muscle activation patterns of tibialis anterior and gastrocnemius lateralis. The large stride-to-stride variability of gait data was handled in an innovative way, processing separately: 1) distinct foot-floor contact patterns, and for each specific foot-floor contact pattern 2) distinct muscle "activation modalities", averaging only across gait cycles with the same number of activations, and obtaining, in both cases, the pattern frequency-of-occurrence. RESULTS At least 2 representative foot-floor contact patterns within each Winters group, and up to 4-5 distinct muscle activation patterns were documented. CONCLUSIONS It cannot be defined a predominant muscle activation pattern specific for a Winters group. For a correct clinical assessment of a hemiplegic child, it is advisable to record and properly analyze gait signals during a longer period of time (2-3 min), rather than (subjectively) selecting a few "clean" gait cycles, since these cycles may not be representative of the patient's gait.