The aims of the article were to study the reproducibility of latency of multifocal visual evoked potential (mfVEP) recorded using different stimulus presentations and to identify the peak with least variability. Ten normal subjects, aged between 22 and 52 years (mean age 32 ± 8.37 years), participated in the study. All subjects underwent mfVEP testing with pattern reversal and pattern pulse stimulus presentations. The stimulus subtends 26° from fixation and includes 24 segments. Only the vertical channel was recorded on all subjects. Testing was repeated after 1-2 weeks. Only the right eye of all subjects was analysed. Segments with low signal-to-noise ratios (SNR < 1.5) were excluded from analysis. The latencies were analysed to confirm values from the same peak for the two tests. The latency values were then analysed for the start of the response, the first peak and the second peak. The waveforms were reproducible throughout the field. Reproducibility of latency at the "start of the response" was significantly lesser than the first and the second peaks studied, while the reproducibility of latency at the first peak was not statistically different from the second peak for either pattern reversal or pattern pulse stimulation. The latency values were not different between the first and the second sessions for either pattern reversal or pattern pulse stimulation for any of the peaks. The pattern reversal stimulus presentation produced less variability in latency. The first peak is the most reproducible among the three measures in both the stimulus presentation.