This study had, as its aim, to test twelve nebulizers (6 jet 6 ultrasonic) which are used in the treatment of cystic fibrosis. Devices were connected to a respirator in order to mimic the ventilation of a child and of an adult suffering from cystis fibrosis. Three medications: tobramycine, colistine and amiloride were nebulised. The volume of the recommended solution varied between 1.5 and 13 ml according to the manufacturer. During a session often minutes the ultrasonic nebulizer delivered an inhaled volume which was significantly greater than the jet (2.72 +/- 0.98 ml vs 1.22 +/- 0.59 ml, p < 0.0001) for the three drugs. Regarding granulometry, the fraction of particles between 0.5 and 5 mu m, was higher with ultrasonic than with pneumatic nebulizer for tobramycine (67.1 +/- 10.7 vs 55.5 +/- 11.5%, p < 0.001) and amiloride (66.4 +/- 9.2% vs 58.1 +/- 15%, p < 0.05%). The variation of concentration due to nebulisation were independent of the type of apparatus bur influenced by the drug since concentration was increased for tobramycine (+10.5 +/- 18.6%) and amiloride (+13.4 +/- 8/9%). In summary the effective fraction resulting fr om the inhalable fraction, from granulometry and from changes in concentration was significantly greater for ultrasonic than for jet nebulizer (17.3 +/- 6.7% vs 9.7 +/- 9.6%, p < 0.001). This study underlines the great variability of the performance of aerosols generators and therefore the need for an accurate evaluation of nebulizer performances in order to prescribe the best nebulizer/drug association in clinical practice.