Abstract Experiments employing a rodent circling model were conducted to test the predictive capacity of the theory which states that striatonigral γ-aminobutyrate neurones transmit striatal information influencing the animal's locomotion and orientation. In agreement with this proposal, blocking nerve conduction in one substantia nigra with procaine, or nigral γ-aminobutyrate receptors with bicuculline administered stereotaxically, frequently forced rats to move ipsiversively to systemic apomorphine, as though the treatment had impaired striatonigral transmission on that side of the brain. Attempts to reverse the direction of apomorphine circling by stimulating γ-aminobutyrate receptors with muscimol, by facilitating the amino acid's action with flurazepam, or by increasing its synaptic concentration either with a breakdown inhibitor (ethanolamine O-sulphate or 4-amino-hex-5-enoic acid) or an uptake blocker ( cis-1,3-aminocyclohexane carboxylic acid) in one nigra, proved unsuccessful. In fact, ethanolamine O-sulphate, flurazepam and muscimol all gave the appearance of hindering rather than enhancing the passage of striatal-derived motor information through the nigra. Broadly speaking, these drugs gave predictable behavioral responses from the ventromedial thalamus, suggesting they were acting in accordance with known mechanisms. The anomalous behaviour with ethanolamine O-sulphate may be attributed to its elevating γ-aminobutyrate levels in other brain areas, since similar ipsiversive rotations occurred if γ-aminobutyrate catabolism was prevented at a wide variety of extranigral sites. A simple explanation for the paradoxical ipsiversive behaviours produced by intranigral flurazepam or muscimol in combination with systemic or intracerebral injection of dopamine agonists, is that they act via presynaptic receptors to inhibit the release of endogenous γ-aminobutyrate and thereby impede striatonigral outflow ipsilaterally.