Six preparations were considered: three multiple unit dosage forms (micropellets in capsules) (D, E and G) and one matrix tablet (B) were experimental prolonged release formulations, two non-disintegrating tablets (A and C) were commercial products. The in vitro dissolution behaviour of the differing formulations was investigated using the USP XXII paddle apparatus. The in vivo study was effected on a panel of 12 healthy volunteers. The two commercial tablets (A and C) showed mean dissolution time (MDT) of 1.34 and 1.44 h and td of 91 and 92 min, respectively; for prolonged release formulations (B, E, D, and G) MDT ranged between 2.28 and 4.23 h and td between 149 and 291 min. The mean residence time (MRT) was 8.68 and 6.47 h for tablets A and C, respectively; it ranged between 9.62 and 10.24 h for the multiple unit formulations E, D, and G and was 11.27 h for matrix B. Formulation B also showed the higher apparent elimination half-life t1/2 (7.12 h), while apparent t1/2 for all the other formulations were very similar, ranging between 5.04 and 5.28 h. High variability between the various formulations was found for Cmax and AUC values, and no relationships could be established with the type of formulation. An in vitro/in vivo correlation was found for all the formulations examined on the basis of analogous parameters (MDT and MRT); (r = 0.83, p < 0.05). In a few cases the Wagner-Nelson deconvolution method was applied to individual plasma level versus time curves and the corresponding absorption curves were obtained. In these cases the in vitro/in vivo correlation was tested on the basis of the comparison of the in vivo absorption curves with the in vitro dissolution profiles. This was accomplished using the 'Levy's plot' (per cent released versus per cent absorbed) approach and provided further support for the correlation found.