This study used the quasi-experimental type of research to determine the usefulness of an activity-based instruction in improving the speaking skills of the grade ten students in Arayat, Pampanga Purposive sampling technique was used in choosing the 34 participants. The Foreighn Language Classroom Anxiety Scale adopted from Horwitz, Hortwitz and Cope (1986) served as the tool in classifying the participants' perceived level of anxiety in speaking. The extemporaneous speech served as the pretest and posttest, which consisted of 68 question-form topics with the same level of difficulty. The researcher and two inter-raters evaluated the participants' speaking performance using the rubrics adopted from grade ten learner's material tutled Celebrating Diversity through World Literature. To interpret the data, mean, standard deviation and t-test were used. The findings revealed that the participants' perceived level of speaking anxiety mostly fell under the moderate level, wherein the fear of negative evaluation and committing grammatical errors were highlighted. The posttest scores of the participants established significant improvement from the scores both given by the researcher and the inter-raters. The scores as assessed by the researcher and the inter-raters reflected no significant differences; therefore, scores were comparable with one another. The pretest and posttest scores as one group showed a highly significant differences; thus, the activity-based instruction had promising features that positively influence the participants. the participants' pretest and posttest scores as classified based on the level of their anxiety indicated a highly significant difference that could also suggest a positive impact of ABI to the participants.