Lithium has been used to treat depressive illnesses for half a century, yet its biochemical mechanism of action remains obscure. Lithium-induced changes in some cellular ions (e.g. potassium, magnesium) have been described in cardiac Purkinje fibres and brain tissue [ 1, 21. Since blood platelets share many properties with neurones [3], and have previously been used as models for their behaviour in depression (e.g. [4]), we have studied the effect of lithium on washed bovine platelets. Fresh cattle blood was collected at the local abattoir into acid citrate dextrose [5]. The citrated blood was diluted with modified HEPES-buffered Tyrode (b1ood:buffer ratio 3:1, pH 7.4) containing no added Mg2+ or Ca2+, then centrifuged (15 min at 400g at RT) to obtain platelet-rich plasma (PRP). To measure free ionised Ca (Ca2+) or Mg (Mg2+), PRP was incubated (30 min at RT) with 5 mM h a 2 or magfura-2-acetoxymethyl ester/0.025% Pluronic F-127 wlv. Platelets for aggregation or ion measurements were washed by adding the modified Tyrode, centrifuging, and resuspending the pellet to give 0.5 2 x lo8 platelets/ml. Ca and Mg were added back as required followed by a 15 min incubation at 37°C before the start of any experiment. Lithium chloride and bis-dimethylammonium chloride (BDA) were added in the ratios 0:20, 5: 15, or 20:0, to give final Li concentrations of 0 (control), 5 or 20 mM. For the zero Ca2+ experiments, EGTA was added instead of CaClz to give a final concentration of 500 pM. Platelet aggregation was followed using the turbidometric method [6]. For ion studies, platelets were incubated in a SPEX fluorimeter at 37C, with excitation of 340 and 380 nm (sample frequency 0.5 Hz), and emission at 510 nm. The calibration references Rmax, Rmh and B were obtained by adding Triton (0.2% final concentration) to platelets with 1 mM Ca2+, followed by TRIS-EGTA (15 mM final concentration, pH 7.4). Using the equation described by Grynluewicz [7], and assuming a & of 224nM for fura-2, and 1.5 mM for magfbra-2 [8], [CaZ+], and [Mg2+], were calculated. Values shown are means f SEM, and statistical differences were evaluated by Student's paired or unpaired I tests. Dose-dependent responses to thrombin were obtained in the presence and absence of Li (Figure I). When expressed as a percentage of the control value, 5 and 20 mM Li reduced platelet aggregation induced by 2 U/ml thrombin (89.9 f 9.1% (n=5) and 74.2 f 9.6% (n=6), p < 0.05 and p < 0.02 respectively). Resting platelet [Ca2+], and [Mg2+], were 369f62nM (n=12) and (n=7)