Study of geographical variations in the rate of surgery for glue ear reveals striking differences both between English health regions and between health districts. There are two-fold differences between regions and up to seven-fold differences between districts. Analysis of these differences at different levels of population aggregation reveals that professional uncertainty about the indications and value of surgery for this condition is the major factor responsible. In addition, the availability of otolaryngologists influences the surgical rate at district though not at regional level. The method used in this study is unable to determine the part played by geographical variation in patient expectation and morbidity rate. In contrast to the lack of consensus among otolaryngologists regarding the use of surgery for glue ear, there appears to be a high degree of agreement as regards tonsillectomy. The latter has coincided with the end of the decline in the tonsillectomy rate in the late 1970s.