Thermodynamic algorithms can be used to predict RNA secondary structure from sequence. In order to predict the stability of RNA duplexes containing 1x2 internal loops, current algorithms rely on a minimal data set and several approximations. A database of 955 RNA secondary structures was compiled and searched to identify the most common 1x2 loops in nature. Thermodynamic parameters derived from optical melting studies are reported for a series of 24 RNA oligoribonucleotides containing naturally occurring 1x2 loops. This data set doubles the number of 1x2 loops previously studied and, along with previously published data (Schroeder et al., 1996), provides thermodynamic data for the 25 most frequently occurring 1x2 loops in nature. A new model to predict the stability of RNA duplexes containing unmeasured 1x2 loops is derived. Unmeasured 1x2 loops are approximated by DGš37,loop = DGš37,1x2 initiation + DGš37,AU/GU + DGš37,GA/GG + DGš37,UU. Here, DGš37,loop initiation is the free energy of initiation for a 1x2 loop (2.3 kcal/mol), DGš37,AU/GU is a penalty for replacing a closing G-C pair with an A-U or G-U pair (0.6 kcal/mol), DGš37,GA/GG is a bonus for a potential G∙A or G∙G pair in the loop (-1.3 kcal/mol), and DGš37,UU is a bonus for a potential U∙U pair in the loop (-1.0 kcal/mol). A table of thermodynamics for frequently occurring 1x2 loops and a new model to predict the thermodynamics of unmeasured 1x2 loops are likely to improve free energy calculations for duplexes containing 1x2 loops and, therefore, also improve secondary structure prediction from sequence.