The Inquérito à estrutura das explorações agrícolas is a National Agricultural Survey conducted by the Portuguese Statistical Office (INE) and based on a random stratified sample. Inference is made using the Horvitz-Thompson estimator. Estimates of this survey are published at Região Agrária level, as it is considered the lowest level of aggregation for witch the estimates present adequate precision.
This article results from a research project aiming to understand if inference can be made at a lower level of aggregation (as NUTIII), and to propose possible alternative estimators to Horvitz-Thompson. Several approximately unbiased estimators for area estimation are proposed.

The proposed estimators are special forms of the regression estimator. They can combine sample data with auxiliary data arising from agricultural census or administrative sources. Approximate variances of the proposed estimators are deducted under the stratified sampling design and their estimates are produced using data from the 1993 survey. The
 aggregation of estimates is addressed and proposed transformations to the estimators that guaranty their consistency with the Horvitz-Thompson estimate at a higher level of aggregation. The proposed estimators are also compared in terms of precision and recommendation regarding their use for domain estimation proposes is made.

keywords: area estimation, domain estimation, gap allocation, agricultural survey.