In this paper, a new direct method of calculating the first-order sensitivity coefficients using sparse matrix technology to chemical kinetics is presented. The Gear type procedure is used to integrate a model equation and its coupled auxiliary sensitivity coefficient equations. Because the Jacobian matrix of the model equation is the same as that of the sensitivity coefficient equation with respect to a parameter, it is only necessary to triangularize the matrix related to the Jacobian matrix of the model equation. The FORTRAN subroutines of the model equation, the sensitivity coefficient equations, and their Jacobian analytical expressions are generated automatically from chemical mechanism. This method greatly increases the efficiency of computation by taking advantage of the fact that the auxiliary equations for different sensitivity coefficients are linear and quite similar. Two sets of chemical reactions are used to illustrate this approach: oxidation of formaldehyde in the presence of carbon monoxide and photo-oxidation of dimethyl sulfide. The accuracy and computational efficiency of the new direct method is demonstrated by comparing the results from the new direct method and from the indirect method.