SummaryThe beneficial effect of bromocriptine in fluctuating Parkinson's disease (PD) under chronic levodopa therapy has been reported to be lost in the long term in most patients. We attempted slow bromocriptine discontinuation in nine PD patients to whom this drug have been added to levodopa because of fluctuating motor responses but who appeared to cease benefitting from this association after an average of 4.6 years (range, 1.3–8 years) of sustained bromocriptine therapy. Bromocriptine was discontinued by gradually tapering off the dosage (mean, 8.5 mg/week) while leaving levodopa medication unchanged. The percentage time spent “off” during waking hours increased in all patients (average, 30.2 ± 18.5% versus 65.6 ± 82.6%, p < 0.001) and in three patients the “on-off” pattern shifted from random fluctuations to disabling “off” periods 23–26 times longer than before discontinuation. Sustained bromocriptine therapy does not prevent the continuous deterioration of motor fluctuations. However, we believe that this drug remains partially effective over the long term and should not be discontinued in PD patients with uncontrolled fluctuations, even though their condition may actually appear to worsen as compared to when bromocriptine therapy was first started.