Normalization of surface electromyography (sEMG) signal amplitude is considered as a necessary operation to enable comparable data on different muscles, individuals, and sessions. Previous studies usually suggested using the maximal contraction normalization procedure. However, that procedure might not always be possible or the best method in some sEMG studies. The purpose of this study is therefore twofold. The first is to investigate reliability of two different constant load normalization procedures (with and without feedback) at different constant-force submaximal contractions. The second is to investigate correlation of normalization factors obtained from maximal voluntary and standardized submaximal tasks. 18 young healthy participants took part in the study. Subjects performed three muscle contraction tasks, namely, (i) maximal voluntary contraction (MVC) task: isometric maximal contraction of biceps brachii muscle, (ii) force matching task (FM): matching 2.5 kg, 5.0 kg, 7.5 kg and 10.0 kg force with visual feedback, and (iii) load holding (LH) task: holding 2.5 kg, 5.0 kg, 7.5 kg and 10.0 kg weights without visual feedback. sEMG amplitude normalization factors were examined for three tasks. The results of the study suggested that the reliability of sEMG amplitude normalization factors from FM and LH tasks for four target forces or loads were high (intraclass correlation (ICC): 0.863-0.958) to very high (ICC: 0.970-0.995). Due to some limitations of the MVC maximal contraction normalization procedure, normalization to the maximal might not always be possible or the best method for some sEMG studies. In such cases, submaximal isometric load holding tasks could be an alternative to the MVC task for biceps brachii muscle.