Given an undirected connected graph G, the profile minimization problem (PMP) is to place the vertices of G in a linear layout (labeling) in such a way that the sum of profiles of the vertices in G is minimized, where the profile of a vertex is the difference of its labeling with the labeling of its left most neighbor in the layout. It is an NP-complete problem and has applications in various areas such as numerical analysis, fingerprinting, and information retrieval. In this paper, we design a tabuembedded simulated annealing algorithm for profile reduction (TSAPR) for PMP which uses a well-known spectral sequencing method to generate an initial solution. An efficient technique is employed to compute the profile of a neighbor of a solution. The experiments are conducted on different classes of graphs such as T 4 -trees, tensor product of graphs, complete bipartite graphs, triangulated triangle graphs, and a subset of Harwell-Boeing graphs. The computational results demonstrate an improvement in the existing results by TSAPR in most of the cases.