Agreement with coordinated subjects in Slavic languages has recently seen a rapid increase in theoretical and experimental approaches, contributing to a wider theoretical discussion on the locus of agreement in grammar (cf. Marušič, Nevins, and Saksida 2007; Bošković 2009; Marušič, Nevins, and Badecker 2015). This paper revisits the theoretical predictions proposed for conjunction agreement in a group of South Slavic languages, with a special focus on gender agreement. The paper is based on two experiments involving speakers of Bosnian/Croatian/Serbian (BCS) and Slovenian (Sln). Experiment 1 is an elicited production experiment investigating preverbal-conjunct agreement, while Experiment 2 investigates postverbal-conjunct agreement. The data provide experimental evidence discriminating between syntax proper and distributed-agreement models in terms of their ability to account for preverbal highest-conjunct agreement and present a theoretical mechanism for the distinction between default agreement (which has a fixed number and gender, independent of the value of each conjunct) and resolved agreement (which computes number and gender based on the values of each conjunct and must resolve potential conflicts). Focusing on the variability in the gender-agreement ratio across nine combinations, the experimental results for BCS and Sln morphosyntax challenge the notion of gender markedness that is generally posited for South Slavic languages.