The phosphorylation of a peptide substrate by the catalytic subunit of cAMP-dependent protein kinase was monitored over short time periods (2-1000 ms) using a rapid quench flow mixing device and a radioactive assay. The production of phosphokemptide [LRRAS(P)LG] as a function of time is characterized by a rapid "burst" phase (250 s-1) followed by a slower, linear phase (L/[E]t = 21 s-1) at 100 microM Kemptide. The amplitude of this "burst" phase varies linearly with the enzyme concentration and represents approximately 100% of the total enzyme concentration, indicating that the "burst" phase is not due to product inhibition. The observed rate constants for the "burst" and linear phases and the "burst" amplitude vary hyperbolically with the substrate concentration. From these dependencies, a maximum "burst" rate constant of 500 +/- 60 s-1 and a Km and Kd for Kemptide of 4.9 +/- 1.4 and 200 +/- 60 microM were determined. The kcat and Km data extracted from the linear portion of the rapid quench flow transients are indistinguishable from those obtained by standard steady-state kinetic analyses using low catalytic subunit concentrations and a spectrophotometric, coupled enzyme assay. Both rate constants for the "burst" and linear phases decreased in the presence of Mn2+. The data imply that the phosphorylation of Kemptide by the catalytic subunit occurs by a mechanism in which the substrate is loosely bound, is rapidly phosphorylated at the active site, and is released at a steady-state rate that is likely controlled by the dissociation rate constant for ADP. The combined pre-steady-state kinetic data establish a comprehensive, kinetic mechanism that predicts all the steady-state kinetic and viscosometric data. This study represents the first chemical observation and characterization of phosphoryl transfer at the active site of a protein kinase and will be useful for further structure-function studies on this and other protein kinases.