The striatum plays a central role in motor control, yet how it dynamically represents variables such as locomotion speed, particularly across varying behavioral contexts, remains incompletely understood. Here, we investigated striatal encoding of locomotion speed in rats performing an automated T-maze task. We found that the activity of most (78%) analyzed striatal neurons- referred to as speed cells-was robustly correlated, either positively or negatively, with locomotion speed. This population included both putative medium spiny neurons (MSNs; 74%) and fast-spiking interneurons (FSIs; 82%). Speed-related activity was remarkably stable, showing no significant influence of elapsed time, cue type, spatial choice, or trial outcome. Additionally, positively correlated MSNs tended to precede speed changes, while positively correlated FSI activity typically followed, as did negatively correlated neurons for both types. This suggests distinct roles for different striatal cells in movement modulation. Speed cells exhibited strong modulation at movement onset and offset, yet also maintained correlations with speed throughout locomotion bouts. Finally, the firing rates of speed cells reliably predicted locomotion speed, outperforming non-speed cells and chance levels; decoding accuracy further improved when data from multiple neurons were combined, consistent with a population code. Together, these results demonstrate a robust, context-independent representation of locomotion speed in the rat striatum, driven by diverse cell types, and extends previous findings to a task with greater cognitive demands.