BACKGROUND: Otitis media with effusion (OME), or 'glue ear', is very common in children, especially between the ages of one and three years with a prevalence of 10% to 30% and a cumulative incidence of 80% at the age of four years. OME is defined as middle ear effusion without signs or symptoms of an acute infection. OME may occur as a primary disorder or as a sequel to acute otitis media. The functional effect of OME is a conductive hearing level of about 25 to 30 dB associated with fluid in the middle ear. Both the high incidence and the high rate of spontaneous resolution suggest that the presence of OME is a natural phenomenon, its presence at some stage in childhood being a normal finding. Notwithstanding this, some children with OME may go on to develop chronic otitis media with structural changes (tympanic membrane retraction pockets, erosion of portions of the ossicular chain and cholesteatoma), language delays and behavioural problems. It remains uncertain whether or not any of these findings are direct consequences of OME. The most common medical treatment options include the use of decongestants, mucolytics, steroids, antihistamines and antibiotics. The effectiveness of these therapies has not been established. Surgical treatment options include grommet (ventilation or tympanostomy tube) insertion, adenoidectomy or both. Opinions regarding the risks and benefits of grommet insertion vary greatly. The management of OME therefore remains controversial. OBJECTIVES: To assess the effectiveness of grommet insertion compared with myringotomy or non-surgical treatment in children with OME. The outcomes studied were (i) hearing level, (ii) duration of middle ear effusion, (iii) well-being (quality of life) and (iv) prevention of developmental sequelae possibly attributable to the hearing loss (for example, impairment in impressive and expressive language development (measured using standardised tests), verbal intelligence, and behaviour). SEARCH STRATEGY: We searched the Cochrane Central Register of Controlled Trials (CENTRAL) (The Cochrane Library Issue 1, 2003), MEDLINE (1966 to 2003), EMBASE (1973 to 2003) and reference lists of all identified studies. The date of the last systematic search was March 2003, and personal non-systematic searches have been performed up to August 2004. SELECTION CRITERIA: Randomised controlled trials (RCTs) evaluating the effect of grommets on hearing, duration of effusion, development of language, cognition, behaviour or quality of life. Only studies using common types of grommets (mean function time of 6 to 12 months) were included. DATA COLLECTION AND ANALYSIS: Data from studies were extracted by two reviewers and checked by the other reviewers. MAIN RESULTS: Children treated with grommets spent 32% less time (95% confidence interval (CI) 17% to 48%) with effusion during the first year of follow-up. Treatment with grommets improved hearing levels, especially during the first six months. In the randomised controlled trials that studied the effect of grommet insertion alone, the mean hearing levels improved by around 9 dB (95% CI 4 dB to 14 dB) after the first six months, and 6 dB (95% CI 3 dB to 9 dB) after 12 months. In the randomised controlled trials that studied the combined effect of grommets and adenoidectomy, the additional effect of the grommets on hearing levels was improvement by 3 to 4 dB (95% CI 2 dB to 5 dB) at six months and about 1 to 2 dB (95% CI 0 dB to 3 dB) at 12 months. Ears treated with grommets had an additional risk for tympanosclerosis of 0.33 (95% CI 0.21 to 0.45) one to five years later. In otherwise healthy children with long-standing OME and hearing loss, early insertion of grommets had no effect on language development or cognition. One randomised controlled trial in children with OME more than nine months, hearing loss and disruptions to speech, language, learning or behaviour showed a very marginal effect of grommets on comprehensive language. AUTHORS' CONCLUSIONS: The benefits of grommets in children appear small. The effect of grommets on hearing diminished during the first year. Potentially adverse effects on the tympanic membrane are common after grommet insertion. Therefore an initial period of watchful waiting seems to be an appropriate management strategy for most children with OME. As no evidence is yet available for the subgroups of children with speech or language delays, behavioural and learning problems or children with defined clinical syndromes (generally excluded from the primary studies included in this review), the clinician will need to make decisions regarding treatment for such children based on other evidence and indications of disability related to hearing impairment. This review does not resolve the discrepancy between parental and clinical observation of a beneficial treatment effect and the results in the reviewed RCT showing only a short-term effect on hearing and virtually no effect on development. Is the perceived, often dramatic, effect of grommets only a short-term one? Are some children more sensitive to OME-related hearing loss than others? If so, how do we identify them?Further research should focus upon indications. Studies should use sufficiently large sample sizes to show significant interactions. There is a need to determine the most suitable variables and appropriate "softer" outcomes to be the subject of these interaction tests. Interesting options include measures of speech-in-noise and binaural hearing. The generally modest results in the trials which are included in this review should make it easier to justify randomisation of more severely affected and higher-risk children in appropriately constructed trials. Randomised controlled trials are necessary in these children before more detailed conclusions about the effectiveness of grommets can be drawn.