Although there are numerous publications reporting eradication results, the general picture is confused by the bewildering multiplicity of treatment schedules employed by the various workers. The over-riding need now is for large scale trials, and more especially for direct comparisons of different treatment regimens in the same populations of patients. Such data are entirely absent from the literature at present. Standardization of definitions and of methodology pertaining to diagnosis of eradication, recording of side effects, measurement of compliance and determination of recurrence or of reinfection, is badly needed. As the definition of eradication remains arbitrary, it is important to include genome fingerprinting techniques in the long-term follow-up for recurrence, so that the question of reinfection versus recrudescence can be examined (Bell et al, 1993b; Xia et al, 1994). Because of the wide differences in the agents used in H. pylori eradication therapies, proper double-blinding of treatment trials remains a difficult problem. This can be dealt with to some extent by ensuring that the interpretation of tests for H. pylori eradication is performed by personnel unaware of the clinical details. Review of the existing data on eradication of H. pylori indicates that clinically useful results can be achieved in some 70 to 95% of patients, on an intention to treat basis. Compliance, side effects and resistance to metronidazole remain the limiting factors. Efficacy, freedom from side effects, simplicity and low cost will determine the success of any regimen in the future. At present, it is not possible to make firm recommendations in favour of one regimen over another, but it seems reasonable to forecast that dual therapies consisting of a PPI and an antibiotic will receive much attention. Preparations consisting of an H2RA associated with a bismuth compound, which are used together with an antibiotic are an interesting approach. Compliance should be as good as with a normal dual therapy and the eradication results look promising (Wyeth et al, 1994; Webb et al, 1994). The advantages of dual therapies that include a PPI lie in their simplicity, in not relying on imidazole for their anti-H. pylori effect but on the profound inhibition of acid output produced by the PPI. Thus PPI based dual therapy can probably evoke better compliance than the more complicated regimens. The use of PPIs has other advantages in addition to decreasing the MIC90 of the antibiotic combined with it. This is because administration of a powerful inhibitor of gastric acid secretion, such as a PPI, will aid the rapid healing of an ulcer crater and will rapidly relieve the symptoms of peptic ulceration. Gastrin releasing peptide-stimulated acid secretion is raised in duodenal ulcer patients to approximately sixfold over control levels according to El-Omar et al (1993b), and although it returns to normal following the eradication of H. pylori, this process takes time to become effective (El-Omar et al, 1993a). Suppression of acid output provides an immediate therapeutic shield, while the decrease in inflammation and acid output secondary to H. pylori eradication can be established. The most widespread resistance to antibiotics exhibited by H. pylori is with respect to imidazoles. The prevalence of metronidazole resistance is widespread in the emergent countries (Glupczynski et al, 1990), but it is also appreciable in the West, especially in women, who may have been given metronidazole in the treatment of pelvic infections (Rautelin et al, 1992; Banatvala et al, 1994). Moreover, H. pylori becomes resistant to metronidazole very easily and often as a result of treatment which includes an imidazole compound (Malfertheiner, 1993; Banatavala et al, 1994). On the other hand, H. pylori resistance to macrolides is not widespread and does not develop easily during their administration. It is difficult to forecast which antibiotic will be the most widely used agent