Universalities and intriguing analogies in the statistics of avalanches are revealed for three physical systems defined on largely different length and energy scales. Earthquakes induced by tectonic scale dynamics, micro-scale level quakes observed from slipping crystallographic planes in metals and a one-dimensional, room-scale spring-block type Burridge-Knopoff model is studied from similar statistical viewpoints. The validity of the Gutenberg-Richter law for the probability density of the energies dissipated in the avalanches is proven for all three systems. By analysing data for three different seismic zones and performing acoustic detection for different Zn samples under deformation, universality for the involved scaling exponent is revealed. With proper parameter choices the 1D Burridge-Knopoff model is able to reproduce the same scaling law. The recurrence times of earthquakes and micro-quakes with magnitudes above a given threshold present again similar distributions and striking quantitative similarities. However, the 1D Burridge-Knopoff model cannot account for the correlations observed in such statistics. Univerzalnosti i intrigantne analogije u statistici lavina otkrivene su za tri fizičkasustava definirana na uvelike različitim duljinama i energijskim skalama. Potresi uzroko-va ni dinamikom na tektonskoj skali, mikro-potresi koji nastaju na klizećim kristalografskimravnina u metalima i jednodimenzionalni Burridge-Knopoffov model opruga i blokova na skali sobe proučeni su sa sličnih statističkih stajališta. Valjanost Gutenberg-Richteroverelacije za gustoću vjerojatnosti energija disipirane u lavinama dokazana je za sva tri sustava. Analizom podataka za tri različita seizmički aktivna područja i detekcijom akustičkih valova za različite uzorke Zn pod deformacijom, otkrivena je univerzalnost za uključeni eksponent skaliranja. S pravilnim izborom parametara 1D Burridge-Knopoffovmodel može reproducirati isti zakon skaliranja. Vremena ponavljanja potresa i mikro-potresa s magnitudama iznad zadanog praga opet predstavljaju slične distribucije i zapanjujuće kvantitativne sličnosti. Međutim, 1D Burridge-Knopoffov model ne može objasniti korelacije opažene u takvim statistikama.