For the ARMA(p1, p2) model: Xt – θ1Xt-1 - … - θp1Xt-p1 = ?t – α1?t-1 - …- αp2?t-p2 where { ?t} are distributed as i.i.d.N(0,σ2), setting p = p1 + p2 we consider the problem of estimation of the px 1 parameter vector ς=(θ’,α’)’ where θ’ = (θ1,….,θp1) and α’ = (α1,...,αp2), when it is suspected that ς may belong to the subspace Hς = h, where both the m x p matrix H and the m x 1 vector h consist of constants. Five estimators are defined and their asymptotic distributional bias, MSE matrices, and risk expressions are obtained and compared.