This paper describes a bench-scale study dealing with the removal of heavy metals by electrokinetic (EK) remediation from sediment of the Great Backa Canal (Vojvodina, Republic of Serbia), with an emphasis on the dependence of removal efficacies on the physicochemical states of the heavy metals and sediment chemistry. Sediment samples were spiked with the following heavy metals (mg kg(-1)): Zn 4400, Ni 900, Cu 1140 and Cd 57. In addition to determining the pseudo-total metal content in the contaminated sediment before and after EK treatment, BCR sequential extraction was also performed to examine the distribution of the contaminants in the sediment. Conventional EK remediation (EXP I) was ineffective in removing the heavy metals investigated, so two enhanced processes were developed. In both these processes, the mass of treated sediment was reduced to avoid the presence of inactive electric field areas in the sediment and increase current density. The first enhanced experiment (EXP II) used acetic acid (HAc) solution (pH 2.9) as an anolyte. Combined with the smaller sediment mass, this resulted in an increase in overall removal efficacies (9% for Zn, 15% for Ni, 10% for Cu and 15% for Cd). The second enhanced experiment (EXP III), as well as using HAc solution as an anolyte, made use of a cation exchange membrane in the cathodic chamber to minimize pH changes in the region adjacent to the cathode, which negatively influenced the removal of some heavy metals. However, no improvement in removal efficacy was achieved in EXP III. Since the redox potential of the sediment drops during the EK process, metals removal is limited by the formation of their sulfides. In conclusion, the removal of heavy metals by EK remediation is governed by a complex interplay of the complexation, precipitation and reduction processes, and the difficulties encountered in their optimization can explain the unsatisfactory effectiveness achieved by the described remediation procedure. Improved understanding of the behavior of metal ions during EK treatment can be useful in predicting and enhancing the efficacy of the process.