There is a great interest in experiments in the contemporary city. Many diverse initiatives use the term to describe their efforts of questioning existing configurations or proposing new ideas. In the wider literature on urban change, experimentation is frequently suggested as a method for bringing into being the uncertain and hopeful alternatives of critical or insurgent utopias. Yet there is little explicit discussion of what experimentation actually means in these contexts, which fuels concerns over an empty buzzword. Some useful ideas have started to emerge from discussions of grassroots innovations and urban laboratories, which highlight local relevance and strategic visibility as important markers of experimentation. Yet they also show the need for wider conceptual and empirical work on experiments that are explicitly urban and alternative in their outlook. This research takes up this challenge by interrogating the underlying notion of experimentation, tracing its development from a scientific method towards an adaptive and action-oriented social practice of knowledge-making. This extension highlights the importance of place and the involvement of heterogeneous, more-than-human collectives in this process. It also reveals collective experiments as relational practices of negotiation that create knowledge through surprise and adjustment, and which can best be conceptualised as socio-material assemblages.To understand how urban grassroots alternatives assemble their experimentality, the research works with three case studies that describe themselves as experiments: the autonomous town of Christiania in Copenhagen, the eco-squat of Can Masdeu in Barcelona, and the Prinzessinnengarten, an urban garden in Berlin. Using specific constructions, routines and projects for each site, the discussion traces their experimental dimensions through material practices, social dynamics and underlying approaches. It highlights their assembling as processes of balancing, integration and cultivation respectively. Despite these different modes of emergence, the cases indicate some shared experimental features: a focus on learning as knowing-in-practice, an ambiguous relationship of separation from and interconnection with the city, and a clear commitment to publicness. This casts experimental alternatives as distinctive sites of urban learning that make visible alternative modes of dwelling and enable translation through situated adaptation. It also presents them as sources of an urban innovation that is incremental and improvised rather than based on radical novelty. Experimental alternatives extend the promises of urban grassroots interventions by opening up wider avenues of engagement and participation, which suggests experimentation as a useful approach in the work of such initiatives.