Dynamic trade-off theory proposes that firms may deviate from their target capital structure but they will exhibit an adjustment behavior towards that target. Estimating the speed of adjustment (SOA) is an investigation for the joint hypotheses that the target actually exists and that firms adjust toward their target. However, using a single estimated SOA that fits all firms is misleading and cannot be used as evidence for or against the dynamic trade-off theory if the real dynamic behavior is largely heterogeneous. This study finds evidence for this heterogeneity in the SOA for Malaysian firms using system GMM approach. The study finds that firms that are far from the target exhibit faster adjustments than firms close to the target, and firms that are overleveraged exhibit faster adjustment than underleveraged firms. These results are consistent with the dynamic trade-off theory. However, the finding of this study cannot reject other interpretations of capital structure i.e., pecking order or timing theory, since the SOA for some of the firms is very slow and hence, cannot be the first order determinant of capital structure.