CPAP should be considered the first line of treatment in patients with moderate to severe obstructive sleep apnoea. In our centre in Sydney this generally means patients with more than 20 apnoea/hypopnoeas per hour with repeated dips in oxyhaemoglobin saturation and usually some symptomatology. Despite this first line role of nasal CPAP, recent objective studies question whether earlier enthusiastic reports on adherence to CPAP are correct. The role of technical innovations in new CPAP machines in improving usage remains to be tested. The "drop out" rate from physician selection for a CPAP trial to highly compliant user is certainly more than 50% of patients. What happens to these patients? Data from some studies suggest that surgical treatments are used, at least in the USA, but in all probability many of these patients remain untreated. The challenge in the next decade is either to improve CPAP devices to increase usage in this group or to develop other treatment options. The role of intensive inhospital "acclimatisation" to CPAP also has yet to be objectively tested. It is unclear whether "intelligent" CPAP will make huge inroads in increasing the number of patients who accept CPAP trials, prescriptions, or compliance. It will have minimal impact on patients with mask problems or claustrophobia or those who feel that CPAP is inconvenient. There is a high likelihood that it will reduce technologist workload during CPAP titration studies. "Intelligent" CPAP may help to reduce total overnight mouth leakage and therefore reduce nasal side effects. The current expense of developing such devices will mean that they are unlikely to supersede much cheaper standard "one pressure" CPAP machines in the next few years.