The malaria parasite Plasmodium falciparum harbours a relict plastid (termed the apicoplast) that has evolved by secondary endosymbiosis. The apicoplast is surrounded by four membranes, the outermost of which is believed to be part of the endomembrane system. Nuclear-encoded apicoplast proteins have a two-part N-terminal extension that is necessary and sufficient for translocation across these four membranes. The first domain of this N-terminal extension resembles a classical signal peptide and mediates translocation into the secretory pathway, whereas the second domain is homologous to plant chloroplast transit peptides and is required for the remaining steps of apicoplast targeting. We explored the initial, secretory pathway component of this targeting process using green fluorescent reporter protein constructs with modified leaders. We exchanged the apicoplast signal peptide with signal peptides from other secretory proteins and observed correct targeting, demonstrating that apicoplast targeting is initiated at the general secretory pathway of P. falciparum. Furthermore, we demonstrate by immunofluorescent labelling that the apicoplast resides on a small extension of the endoplasmic reticulum (ER) that is separate from the cis-Golgi. To define the position of the apicoplast in the endomembrane pathway in relation to the Golgi we tracked apicoplast protein targeting in the presence of the secretory inhibitor Brefeldin A (BFA), which blocks traffic between the ER and Golgi. We observe apicoplast targeting in the presence of BFA despite clear perturbation of ER to Golgi traffic by the inhibitor, which suggests that the apicoplast resides upstream of the cis-Golgi in the parasite's endomembrane system. The addition of an ER retrieval signal (SDEL) - a sequence recognized by the cis-Golgi protein ERD2 - to the C-terminus of an apicoplast-targeted protein did not markedly affect apicoplast targeting, further demonstrating that the apicoplast is upstream of the Golgi. Apicoplast transit peptides are thus dominant over an ER retention signal. However, when the transit peptide is rendered non-functional (by two point mutations or by complete deletion) SDEL-specific ER retrieval takes over, and the fusion protein is localized to the ER. We speculate either that the apicoplast in P. falciparum resides within the ER directly in the path of the general secretory pathway, or that vesicular trafficking to the apicoplast directly exits the ER.