This paper focuses on the prosodic organization of the Lizu language (Qiangic, Tibeto-Burman) of Western Sichuan, China. Previous research (Chirkova and Chen 2012, 2013; Chirkova 2016) suggests that Lizu has a hybrid prosodic system, combining (a) two lexical tones on monosyllabic words (L and H) and (b) three prosodic patterns with stress-like and tonal characteristics on polysyllabic words and compounds (LH, HH, HL). Following this line of research, the present paper aims to yield specific rules and generalizations to capture (i) the organization of the system and (ii) the exact relationship between stress and tone. The paper proceeds from a study of the phonetic properties of stress and tone in Lizu. Duration, intensity, and f0 were chosen as measurement parameters of stress, given that they are cross-linguistically common correlates of stress (e.g. Fry 1955, 1958; Bondarko et al. 1973; Beckman 1986; Gordon 2004, 2011, 2016; Gordon & Nafi 2012). For the same reason, f0 was chosen as a measurement parameter of tone. The corpus for the study was collected with eight subjects (four female and four male), all native speakers of Lizu (the variety of Kala Township, Muli County, Sichuan Province). The list of targeted words comprised 30 disyllabic items (lexical words), which were presented to the subjects in randomized order in two contexts: (a) in isolation, where the word is equivalent to a phrase, and (b) in the phrasal context. Each of the targeted words was read four times in both contexts. Mean duration, mean intensity, and fundamental frequency (at the 1/3, 2/3, and 3/3 of each vowel) were measured for each vowel with the aid of a Praat script. Resulting data were subjected to statistical analysis (repeated measures ANOVAs pooled over all speakers for mean duration and mean intensity, and over male and female speakers separately for f0 measurements). The results of the study point to a clear binary distinction within the three prosodic patterns on polysyllabic words and compounds: HL versus HH & LH. The former differs from both the latter in intensity and duration (common acoustic correlates of stress), whereas the main inner difference between the latter two patterns pertains to f0 measurements (the crucial acoustic correlate of tone). These findings are further linked to the synchronically productive tone derivation rules in Lizu. I suggest that Lizu has both lexical tone (cued by pitch) and stress (cued by intensity and duration). Stress regularly occurs on the initial syllable in a domain (trochaic). Only the initial, stress-able syllable in the domain may bear tonal contrast. Hence, the two lexical tones (H & L) can only be realized on the initial, stressed syllable, spreading to the remaining syllables to their right. Intonational pitch accent is linked to the stressed syllable, however, in domains beginning with the L tone, a combination of tones on the domain-initial syllable is avoided, so that the intonational pitch accent shifts to the domain final syllable. Lizu stress is always obligatory (all major category words are stressed), while some words may be toneless (and bearing the HL pattern). Theoretical and practical implications of these findings are discussed.