Key words and Phrases: adapttve normal h e a r model; autoregresstv~ model; fundanenta: dentaty of seqtieiitzal analysts; m a z z m u n !tkelthood estzmator: zlrlhozun zrarzabtlzty; very weak erpanszon. ABSTRACT A linear model is considered in which the design variables may be functions of previous responses and /or auxiliary randomisation. T h e model is observed t successive times, where t is a stopping time, and interest lies in estimating the parameters of the model. Approximations are derived for the bias and variance of the maximum likelihood estimators of the parameters a t time t . T h e derivations involve differentiating the fundamental identity of sequential analysis. The accuracy of the approximations is assessed by simulation for a multi-armed clinical trial model proposed by Coad (1995), two autoregressive models and the sequential des~gn of Ford and Silvey (1980). Very weak expansions are used t o justify the approximations. Downloaded by [Columbia University] at 15: