Several factors constrain the implementation of Randomised Controlled Trials (RCTs). To obtain large sample sizes a multicentred multinational trial may be necessary or a long sampling period. The larger the trial the larger is the unit cost. To allow larger sample sizes, shorter sampling periods and lower unit costs, new methods are needed. The Internet and in particular the WWW provides such an opportunity. The WWW can provide global access, fast interaction and automation. A prototype Internet Trials Service (ITS) is currently being tested with a real international clinical trial (the Growth Restriction Intervention Trial--GRIT). The ITS is hosted on a Web server. It provides a series of HTML documents that describe the GRIT protocol. Registered centres may enter patients into the GRIT trial via ITS. Java applets are used to collect trial data before returning the study number and randomisation. ITS assumes all trial data will be intercepted by a sniffer. Therefore no information is sent that could specifically identify a patient, this must be sent later by more secure means. ITS assumes that trial centres can be spoofed. To authenticate the patients entered into the trial and the trial data sent, a regular audit report is sent to each centre by secure means for confirmation. By using Java, a full functional data entry system can be developed that runs locally within any Java enabled browser. It can perform data validation locally and also provide a sophisticated user interface.