Aim of the present paper was to investigate the influence of preparation parameters on the production of lipospheres (LS) for drug delivery. LS composed of triglycerides and monoglycerides were alternatively produced by melt dispersion technique, solvent evaporation or w/o/w double emulsion method. The influence of preparation parameters, such as (a) type and amount of lipids, (b) presence and concentration of surfactants, (c) stirring speed and (d) type of stirrer was studied. In the case of LS prepared by melt dispersion, the use of a lipid composition of cetyl alcohol/cholesterol (2:1, w/w), a 5% (w/w) gelatin solution (50 bloom grades) and 1000 rpm stirring speed resulted in the production of spherical particles, with high percentage of recovery (82%, w/w) a mean diameter of 80 microm and a narrow size distribution. In the case of LS prepared by solvent evaporation, the best results in terms of LS morphology, recovery and size distribution were obtained by the use of a lipid composition of tristearin/monostearate (66:34, w/w), a 1% (w/w) PVA solution, a 750 rpm stirring speed and a 55 mm three-blade turbine rotor. The solvent evaporation method resulted in the production of LS characterised by a smaller size (20 microm mean diameter) but poor mechanical properties with respect to particles with the same composition obtained by the melt dispersion technique (170 microm mean diameter). The use of a combination of lipids and a methacrylic polymer (Eudragit RS 100) overcame this problem, resulting in the production of spherical particles, with a narrower size distribution and good mechanical properties. Two lipophilic drugs, such as retinyl acetate and progesterone, and one hydrophilic drug, sodium cromoglycate (SCG), were encapsulated in LS as model compounds. Lypophilic drugs displayed satisfactory encapsulation efficiencies (over 70% w/w), while SCG was very scarcely encapsulated (about 2% w/w). To solve this drawback, the use of a w/o/w double emulsion strategy was proposed, enabling to increase the encapsulation of SCG up to 50% w/w. Finally, in vitro drug release studies were performed, showing that all drugs were released in a control manner. In particular. the retinyl acetate release efficacy within the first 8 h was 27% of the total amount of the drug, while in the same period, the amount of progesterone released was 63%. With regard to SCG containing LS, the release of the drug was largely influenced by the type of stabiliser of the primary emulsion, in any case the SCG release reached the 100% of the total amount of drug after 5 h from the beginning of the experiment.