The domain of real problems in mechanics often contains curved boundaries. Curved boundaries are often more accurately modelled by curved finite elements than by straight edged elements, as straight sides are perfectly satisfactory if the domain has a polygonal boundary. Because fewer curved elements are required, the effort needed to obtain a solution is usually reduced. If some parts of the boundary are curved, however, elements with at least one curved side are desirable. Our aim in this paper is to consider the triangular element with two straight sides and one curved side. This paper is concerned with explicit formulae for evaluating integrals of rational functions of bivariate polynomial numerators with linear denominators over a unit triangle {0⩽ξ, η⩽1, ξ+η⩽1} in the local parametric two dimensional space ( ξ, η). These integrals arise in finite element formulations of second order linear partial differential equations by use of triangular element with two straight sides and one curved side of quadratic variation which often require relatively large numerical effort to integrate. The curved elements considered here are the four node, six node and ten node triangular elements with one curved side of quadratic variation and the other two sides have straight edges under the isoparametric and subparametric transformations, respectively. We have shown that by use of a method similar to synthetic division, the rational integrals of nth order bivariate polynomial numerator with a linear denominator having ( n+1)( n+2)/2 integrals can be reduced to rational integrals of nth order polynomial numerator in one variate with the same linear denominator having ( n+1) integrals and a simple integral of bivariate polynomial expression containing n( n+1)/2 terms (which is free from denominator), and this amounts to a substantial reduction in the numerical effort for such integrals. The explicit analytical integration formulae (obtained) up to sextic polynomial numerator in bivariates ξ, η due to different element geometry are, for clarity and reference, summarised in tables. Finally three application examples are also considered. For the first example we have used integration formulae derived in all theorems of this paper and explained the detailed computational scheme. For the other two examples computational scheme follows in a similar way. It is observed in the solution of all problems that the displacements and torsional constants are satisfactory when components of all element matrices are calculated by analytical integration formulae derived in this paper. It is also observed that in the calculation of components of element matrices by using numerical integration formulae (e.g., 7-point and 13-point rule) much discrepancy occurs if the element geometry is coarse with a concave curve side. But it is found that the analytical integration technique is always consistent. Therefore the symbolic integration formulae presented in this paper are reliable and may lead to an easy and systematic incorporation of element matrices required in the finite element solution procedure.