In Trinidad and Tobago, the demand for sixth form places to pursue A-level studies continues to outstrip the supply. Typically, school administrators have used performance at the CXC examinations as the main criterion for selection of students to pursue A-level work in a given subject. This paper explores the validity of this practice with respect to the study of chemistry, by presenting the results of a tracer study which determined the relationship between student performance on CXC chemistry during the period 1989-1992 and the corresponding performance of these students on A-level chemistry two years later, 1991-1994. Data were obtained from a sample of 1485 students drawn from the population of public secondary schools in Trinidad and Tobago. The analysis reveals a moderate and statistically significant correlation coefficient of 0.49 between the overall CXC chemistry grade and overall GCE A-level chemistry grade. Higher correlation coefficients were obtained for girls in the sample than for boys. When the data were analysed by school type, the highest correlation was obtained for girls in the traditional girls' schools. Further analysis by multiple regression revealed that 31 percent of the variance in A-level performance was accounted for by the performance on the CXC profile of recall and understanding; a further 10 percent of the variance was due to performance onÂ the CXC enquiry skill profile. The paper argues that, although overall performance on CXC chemistry seems to be acting as a reasonable predictor of performance at A-level chemistry, school administrators need to pay special attention to CXC profile grades. They may also need to consider other factors, especially with respect to the boys who seem capable of showing some improvement in grades in moving from CXC to A-level chemistry.