In the survey described by Al-Mufti et al. (see page 23) blood and hair samples were analysed for total mercury by modified atomic absorption spectrophotometry. The hair samples were divided into 2.5-cm segments and analysed consecutively. The mean blood levels were 34 ng/ml and 7 ng/ml, respectively in those who had and those who had not eaten contaminated bread.Corresponding mean maximum hair mercury values were 136 mug/g and 5 mug/g, respectively. Hair mercury values provided a better discrimination between different categories of exposure than blood mercury values at the time the survey was performed, some months after the end of the outbreak. Those persons who had not eaten contaminated bread but who lived in the area of high exposure had hair mercury values between the values of those who had eaten and those who had not eaten contaminated bread and who lived in the area of low exposure. Sequential estimation of mercury in 2.5-cm segments of hair in women gave information on the period of accumulation of mercury more than 1 year before the time of collection of the samples. It was possible to show an approximate relationship between the maximum hair mercury value and the amount of contaminated bread eaten. The match between the blood mercury level and the severity of poisoning was poor, owing to the length of time that had elapsed between the onset of poisoning and the sampling. With hair mercury, while the group results showed a good relation to the severity of poisoning, in individual cases the match was less good, especially in those persons where an insufficient length of hair was available for analysis. Biological variation in sensitivity to methylmercury was also likely to have been an important factor.