Objectives: several improvements have been made in the design of immunoassayssuch as method of antibody production, labeling, automation and detection technology. The aim ofthe present study was to compare the accuracy and precision of enzyme linked immunofloriscenceassay (ELFA) and electrochemiluminescence assay (ECL), with Elisa for determination of serum TSHlevels. Period: Feb 2014 to Nov 2014. Setting: College of postgraduate studies, University of Al-Neelain, Khartoum, Sudan. Material and Methods: Three commercial control materials low, normaland high levels of TSH, were used for imprecision studies of immunofloriscence assay (ELFA) andelectrochemiluminescence assay (ECL) methods and 120 patients samples including low (20%),normal (50%), and high (30%) TSH levels, were measured by the two methods, and used for methodscomparison. In addition to six assigned prepared pool serum used for linearity evaluation of the twomethods. Results: Inter- and intra-assay CV% for ECL and ELFA was significantly low compared withthe required by the manufacture. (Intraassay CV% for ECL was 2.9%, 2.74%, and 2.55% for low,normal, and high respectively of TSH levels of the control sera. Intraassay CV% for ELFA was 3.95%,3.75%, and 5.73% for low, normal, and high respectively of TSH levels of the control sera. InterassayCV% for ECL was 3.0%, 2.75%, and 2.81% for low, normal, and high respectively of TSH levels ofthe control sera. Intraassay CV% for ELFA was 4.26%, 4.0% , and 5.75% for low, normal , and highrespectively of TSH levels of the control sera. Although the mean TSH levels of the three levels of thecontrol sera measured by ECL & ELFA methods, is significantly difference from assigned TSH meanvalues( low 0.488+/-0.078,normal 6.016+/- 0.952,high 33.651+/-5.39) , but the measured values iswithin the mean range of the assigned means values. ECL; low (0.611 +/- 0.018. p ≤ 0.001), normal(6.6785 +/- 0.183. p ≤ 0.00), high (35.0485 +/- 0.894. p ≤ 0.02). ELFA low (0.50545 +/- 0.020. p ≤0.00) , normal (6.5395 +/- 0.244. p ≤ 0.00), high (31.0350 +/- 1.779. p ≤0.001).The mean TSH levelsof the 120 patients samples measured by ECL & ELFA , is significantly difference , for ECL (15.74+/-1.181 . p ≤ 0.00) when compared with the mean TSH value (14.56 +/- 1.65) of the patients samples.For ELFA method also there is significant difference (13.76 +/- 1.59 , p ≤ 0.00) when comparedwith mean of the assigned TSH values(14.56 +/- 1.65) of the patients samples, but within the targetvalues of the means .The study showed strong relationship between the two TSH levels measuredby ECL( mean 15.74 mIU/L, slope 0.67 , correlation coefficients 0.991, p ≤ 0.00) and by ELFA (mean 13.76 mIU/L, slope 0.54, correlation coefficients 0.995, p ≤ 0.00) with the assigned values(14.56 ) of 120 patients sample .The results illustrates no significant difference of TSH mean level insix prepared pool samples measured by ECL(22.63 mIU/l +/- 1.12 , p ≤ 0.1) and ELFA(19.87mIU/l+/- 1.15 , p ≤ 0.11) when compared with the TSH assigned values(22.54 mIU/l +/-0.96), and withstrong correlation between the two TSH levels measured by ECL(22.63 mIU/l, slope 0.79, correlationcoefficients 0.999, p ≤ 0.00) and ELFA ( mean 19.87mIU, slope 0.68, correlation coefficients 0.985,p ≤ 0.00), with the assigned TSH values (22.54 mIU/l +/-0.96).of the six prepared pool samples.Conclusion: Considerable significant precision and accuracy was manifested by both ECL and ELFAmethods in estimation of TSH levels, but ECL is more precise than ELFA especially in the lower TSHconcentration.