Deuterium abundance re-measurements by Burles & Tytler ( 1998 ; hereafter BT ) yielded D/H = ( 3.3 \pm 0.3 ) \times 10 ^ { -5 } and the robust upper limit D/H < 3.9 \times 10 ^ { -5 } from the z _ { a } = 3.572 system toward Q 1937-1009 . In this new analysis BT adopted multicomponent microturbulent models together with the possibility to vary freely the local continuum level around each \ion H1 line to improve the fit . The procedure failed , however , to fit adequately D Ly- \beta without recourse to an additional H Ly- \alpha contamination at the position of D Ly- \beta . We show that this obstacle may be successfully overcome within the framework of the mesoturbulent model accounting ( in contrast to the microturbulent approximation ) for a correlated structure of the large scale velocity field . Using the same observational data and the original continuum as determined by Tytler et al . ( 1996 ) , we obtained good fits . The one-component mesoturbulent models provide D/H in the range \simeq ( 3.2 - 4.8 ) \times 10 ^ { -5 } and the total hydrogen column density N ( \ion H1 ) \simeq ( 5.6 - 7.0 ) \times 10 ^ { 17 } cm ^ { -2 } . This result is consistent with that found by us from the z _ { a } = 2.504 and z _ { a } = 0.701 systems toward Q 1009+2956 and Q 1718+4807 , respectively . The range for D/H common to all three analyses is D/H \simeq ( 4.1 - 4.6 ) \times 10 ^ { -5 } . This value is consistent with standard big bang nucleosynthesis [ SBBN ] if the baryon-to-photon ratio , \eta , is in the range 4.2 \times 10 ^ { -10 } \lesssim \eta \lesssim 4.6 \times 10 ^ { -10 } , implying 0.0155 \lesssim \Omega _ { b } h ^ { 2 } _ { 100 } \lesssim 0.0167 .